ruby/compile.c

11348 строки
326 KiB
C
Исходник Обычный вид История

/**********************************************************************
compile.c - ruby node tree -> VM instruction sequence
$Author$
created at: 04/01/01 03:42:15 JST
Copyright (C) 2004-2007 Koichi Sasada
**********************************************************************/
#include "ruby/encoding.h"
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
#include "ruby/re.h"
#include "ruby/util.h"
#include "internal.h"
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
#include "encindex.h"
#include <math.h>
#include "vm_core.h"
#include "vm_debug.h"
#include "iseq.h"
#include "insns.inc"
#include "insns_info.inc"
#include "id_table.h"
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
#include "gc.h"
#ifdef HAVE_DLADDR
# include <dlfcn.h>
#endif
#undef RUBY_UNTYPED_DATA_WARNING
#define RUBY_UNTYPED_DATA_WARNING 0
#define FIXNUM_INC(n, i) ((n)+(INT2FIX(i)&~FIXNUM_FLAG))
#define FIXNUM_OR(n, i) ((n)|INT2FIX(i))
typedef struct iseq_link_element {
enum {
ISEQ_ELEMENT_ANCHOR,
ISEQ_ELEMENT_LABEL,
ISEQ_ELEMENT_INSN,
ISEQ_ELEMENT_ADJUST,
2019-08-08 19:32:30 +03:00
ISEQ_ELEMENT_TRACE,
} type;
struct iseq_link_element *next;
struct iseq_link_element *prev;
} LINK_ELEMENT;
typedef struct iseq_link_anchor {
LINK_ELEMENT anchor;
LINK_ELEMENT *last;
} LINK_ANCHOR;
typedef enum {
LABEL_RESCUE_NONE,
LABEL_RESCUE_BEG,
LABEL_RESCUE_END,
LABEL_RESCUE_TYPE_MAX
} LABEL_RESCUE_TYPE;
typedef struct iseq_label_data {
LINK_ELEMENT link;
int label_no;
int position;
int sc_state;
int sp;
int refcnt;
unsigned int set: 1;
unsigned int rescued: 2;
unsigned int unremovable: 1;
} LABEL;
typedef struct iseq_insn_data {
LINK_ELEMENT link;
enum ruby_vminsn_type insn_id;
int operand_size;
int sc_state;
VALUE *operands;
struct {
int line_no;
rb_event_flag_t events;
} insn_info;
} INSN;
typedef struct iseq_adjust_data {
LINK_ELEMENT link;
LABEL *label;
int line_no;
} ADJUST;
typedef struct iseq_trace_data {
LINK_ELEMENT link;
rb_event_flag_t event;
long data;
} TRACE;
struct ensure_range {
LABEL *begin;
LABEL *end;
struct ensure_range *next;
};
struct iseq_compile_data_ensure_node_stack {
const NODE *ensure_node;
struct iseq_compile_data_ensure_node_stack *prev;
struct ensure_range *erange;
};
const ID rb_iseq_shared_exc_local_tbl[] = {idERROR_INFO};
/**
* debug function(macro) interface depend on CPDEBUG
* if it is less than 0, runtime option is in effect.
*
* debug level:
* 0: no debug output
* 1: show node type
* 2: show node important parameters
* ...
* 5: show other parameters
* 10: show every AST array
*/
#ifndef CPDEBUG
#define CPDEBUG 0
#endif
#if CPDEBUG >= 0
#define compile_debug CPDEBUG
#else
#define compile_debug ISEQ_COMPILE_DATA(iseq)->option->debug_level
#endif
#if CPDEBUG
#define compile_debug_print_indent(level) \
ruby_debug_print_indent((level), compile_debug, gl_node_level * 2)
#define debugp(header, value) (void) \
(compile_debug_print_indent(1) && \
ruby_debug_print_value(1, compile_debug, (header), (value)))
#define debugi(header, id) (void) \
(compile_debug_print_indent(1) && \
ruby_debug_print_id(1, compile_debug, (header), (id)))
#define debugp_param(header, value) (void) \
(compile_debug_print_indent(1) && \
ruby_debug_print_value(1, compile_debug, (header), (value)))
#define debugp_verbose(header, value) (void) \
(compile_debug_print_indent(2) && \
ruby_debug_print_value(2, compile_debug, (header), (value)))
#define debugp_verbose_node(header, value) (void) \
(compile_debug_print_indent(10) && \
ruby_debug_print_value(10, compile_debug, (header), (value)))
#define debug_node_start(node) ((void) \
(compile_debug_print_indent(1) && \
(ruby_debug_print_node(1, CPDEBUG, "", (const NODE *)(node)), gl_node_level)), \
gl_node_level++)
#define debug_node_end() gl_node_level --
#else
#define debugi(header, id) ((void)0)
#define debugp(header, value) ((void)0)
#define debugp_verbose(header, value) ((void)0)
#define debugp_verbose_node(header, value) ((void)0)
#define debugp_param(header, value) ((void)0)
#define debug_node_start(node) ((void)0)
#define debug_node_end() ((void)0)
#endif
#if CPDEBUG > 1 || CPDEBUG < 0
#define printf ruby_debug_printf
#define debugs if (compile_debug_print_indent(1)) ruby_debug_printf
#define debug_compile(msg, v) ((void)(compile_debug_print_indent(1) && fputs((msg), stderr)), (v))
#else
#define debugs if(0)printf
#define debug_compile(msg, v) (v)
#endif
#define LVAR_ERRINFO (1)
/* create new label */
#define NEW_LABEL(l) new_label_body(iseq, (l))
#define LABEL_FORMAT "<L%03d>"
2015-07-22 01:52:59 +03:00
#define NEW_ISEQ(node, name, type, line_no) \
new_child_iseq(iseq, (node), rb_fstring(name), 0, (type), (line_no))
2015-07-22 01:52:59 +03:00
#define NEW_CHILD_ISEQ(node, name, type, line_no) \
new_child_iseq(iseq, (node), rb_fstring(name), iseq, (type), (line_no))
/* add instructions */
#define ADD_SEQ(seq1, seq2) \
APPEND_LIST((seq1), (seq2))
/* add an instruction */
#define ADD_INSN(seq, line, insn) \
ADD_ELEM((seq), (LINK_ELEMENT *) new_insn_body(iseq, (line), BIN(insn), 0))
/* insert an instruction before next */
#define INSERT_BEFORE_INSN(next, line, insn) \
ELEM_INSERT_PREV(&(next)->link, (LINK_ELEMENT *) new_insn_body(iseq, (line), BIN(insn), 0))
/* insert an instruction after prev */
#define INSERT_AFTER_INSN(prev, line, insn) \
ELEM_INSERT_NEXT(&(prev)->link, (LINK_ELEMENT *) new_insn_body(iseq, (line), BIN(insn), 0))
/* add an instruction with some operands (1, 2, 3, 5) */
#define ADD_INSN1(seq, line, insn, op1) \
ADD_ELEM((seq), (LINK_ELEMENT *) \
new_insn_body(iseq, (line), BIN(insn), 1, (VALUE)(op1)))
/* insert an instruction with some operands (1, 2, 3, 5) before next */
#define INSERT_BEFORE_INSN1(next, line, insn, op1) \
ELEM_INSERT_PREV(&(next)->link, (LINK_ELEMENT *) \
new_insn_body(iseq, (line), BIN(insn), 1, (VALUE)(op1)))
/* insert an instruction with some operands (1, 2, 3, 5) after prev */
#define INSERT_AFTER_INSN1(prev, line, insn, op1) \
ELEM_INSERT_NEXT(&(prev)->link, (LINK_ELEMENT *) \
new_insn_body(iseq, (line), BIN(insn), 1, (VALUE)(op1)))
#define LABEL_REF(label) ((label)->refcnt++)
/* add an instruction with label operand (alias of ADD_INSN1) */
#define ADD_INSNL(seq, line, insn, label) (ADD_INSN1(seq, line, insn, label), LABEL_REF(label))
#define ADD_INSN2(seq, line, insn, op1, op2) \
ADD_ELEM((seq), (LINK_ELEMENT *) \
new_insn_body(iseq, (line), BIN(insn), 2, (VALUE)(op1), (VALUE)(op2)))
#define ADD_INSN3(seq, line, insn, op1, op2, op3) \
ADD_ELEM((seq), (LINK_ELEMENT *) \
new_insn_body(iseq, (line), BIN(insn), 3, (VALUE)(op1), (VALUE)(op2), (VALUE)(op3)))
/* Specific Insn factory */
#define ADD_SEND(seq, line, id, argc) \
2015-07-22 01:52:59 +03:00
ADD_SEND_R((seq), (line), (id), (argc), NULL, (VALUE)INT2FIX(0), NULL)
* rewrite method/block parameter fitting logic to optimize keyword arguments/parameters and a splat argument. [Feature #10440] (Details are described in this ticket) Most of complex part is moved to vm_args.c. Now, ISeq#to_a does not catch up new instruction format. * vm_core.h: change iseq data structures. * introduce rb_call_info_kw_arg_t to represent keyword arguments. * add rb_call_info_t::kw_arg. * rename rb_iseq_t::arg_post_len to rb_iseq_t::arg_post_num. * rename rb_iseq_t::arg_keywords to arg_keyword_num. * rename rb_iseq_t::arg_keyword to rb_iseq_t::arg_keyword_bits. to represent keyword bitmap parameter index. This bitmap parameter shows that which keyword parameters are given or not given (0 for given). It is refered by `checkkeyword' instruction described bellow. * rename rb_iseq_t::arg_keyword_check to rb_iseq_t::arg_keyword_rest to represent keyword rest parameter index. * add rb_iseq_t::arg_keyword_default_values to represent default keyword values. * rename VM_CALL_ARGS_SKIP_SETUP to VM_CALL_ARGS_SIMPLE to represent (ci->flag & (SPLAT|BLOCKARG)) && ci->blockiseq == NULL && ci->kw_arg == NULL. * vm_insnhelper.c, vm_args.c: rewrite with refactoring. * rewrite splat argument code. * rewrite keyword arguments/parameters code. * merge method and block parameter fitting code into one code base. * vm.c, vm_eval.c: catch up these changes. * compile.c (new_callinfo): callinfo requires kw_arg parameter. * compile.c (compile_array_): check the last argument Hash object or not. If Hash object and all keys are Symbol literals, they are compiled to keyword arguments. * insns.def (checkkeyword): add new instruction. This instruction check the availability of corresponding keyword. For example, a method "def foo k1: 'v1'; end" is cimpiled to the following instructions. 0000 checkkeyword 2, 0 # check k1 is given. 0003 branchif 9 # if given, jump to address #9 0005 putstring "v1" 0007 setlocal_OP__WC__0 3 # k1 = 'v1' 0009 trace 8 0011 putnil 0012 trace 16 0014 leave * insns.def (opt_send_simple): removed and add new instruction "opt_send_without_block". * parse.y (new_args_tail_gen): reorder variables. Before this patch, a method "def foo(k1: 1, kr1:, k2: 2, **krest, &b)" has parameter variables "k1, kr1, k2, &b, internal_id, krest", but this patch reorders to "kr1, k1, k2, internal_id, krest, &b". (locate a block variable at last) * parse.y (vtable_pop): added. This function remove latest `n' variables from vtable. * iseq.c: catch up iseq data changes. * proc.c: ditto. * class.c (keyword_error): export as rb_keyword_error(). * common.mk: depend vm_args.c for vm.o. * hash.c (rb_hash_has_key): export. * internal.h: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@48239 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2014-11-02 21:02:55 +03:00
#define ADD_SEND_WITH_FLAG(seq, line, id, argc, flag) \
2015-07-22 01:52:59 +03:00
ADD_SEND_R((seq), (line), (id), (argc), NULL, (VALUE)(flag), NULL)
* rewrite method/block parameter fitting logic to optimize keyword arguments/parameters and a splat argument. [Feature #10440] (Details are described in this ticket) Most of complex part is moved to vm_args.c. Now, ISeq#to_a does not catch up new instruction format. * vm_core.h: change iseq data structures. * introduce rb_call_info_kw_arg_t to represent keyword arguments. * add rb_call_info_t::kw_arg. * rename rb_iseq_t::arg_post_len to rb_iseq_t::arg_post_num. * rename rb_iseq_t::arg_keywords to arg_keyword_num. * rename rb_iseq_t::arg_keyword to rb_iseq_t::arg_keyword_bits. to represent keyword bitmap parameter index. This bitmap parameter shows that which keyword parameters are given or not given (0 for given). It is refered by `checkkeyword' instruction described bellow. * rename rb_iseq_t::arg_keyword_check to rb_iseq_t::arg_keyword_rest to represent keyword rest parameter index. * add rb_iseq_t::arg_keyword_default_values to represent default keyword values. * rename VM_CALL_ARGS_SKIP_SETUP to VM_CALL_ARGS_SIMPLE to represent (ci->flag & (SPLAT|BLOCKARG)) && ci->blockiseq == NULL && ci->kw_arg == NULL. * vm_insnhelper.c, vm_args.c: rewrite with refactoring. * rewrite splat argument code. * rewrite keyword arguments/parameters code. * merge method and block parameter fitting code into one code base. * vm.c, vm_eval.c: catch up these changes. * compile.c (new_callinfo): callinfo requires kw_arg parameter. * compile.c (compile_array_): check the last argument Hash object or not. If Hash object and all keys are Symbol literals, they are compiled to keyword arguments. * insns.def (checkkeyword): add new instruction. This instruction check the availability of corresponding keyword. For example, a method "def foo k1: 'v1'; end" is cimpiled to the following instructions. 0000 checkkeyword 2, 0 # check k1 is given. 0003 branchif 9 # if given, jump to address #9 0005 putstring "v1" 0007 setlocal_OP__WC__0 3 # k1 = 'v1' 0009 trace 8 0011 putnil 0012 trace 16 0014 leave * insns.def (opt_send_simple): removed and add new instruction "opt_send_without_block". * parse.y (new_args_tail_gen): reorder variables. Before this patch, a method "def foo(k1: 1, kr1:, k2: 2, **krest, &b)" has parameter variables "k1, kr1, k2, &b, internal_id, krest", but this patch reorders to "kr1, k1, k2, internal_id, krest, &b". (locate a block variable at last) * parse.y (vtable_pop): added. This function remove latest `n' variables from vtable. * iseq.c: catch up iseq data changes. * proc.c: ditto. * class.c (keyword_error): export as rb_keyword_error(). * common.mk: depend vm_args.c for vm.o. * hash.c (rb_hash_has_key): export. * internal.h: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@48239 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2014-11-02 21:02:55 +03:00
#define ADD_SEND_WITH_BLOCK(seq, line, id, argc, block) \
2015-07-22 01:52:59 +03:00
ADD_SEND_R((seq), (line), (id), (argc), (block), (VALUE)INT2FIX(0), NULL)
#define ADD_CALL_RECEIVER(seq, line) \
ADD_INSN((seq), (line), putself)
#define ADD_CALL(seq, line, id, argc) \
2015-07-22 01:52:59 +03:00
ADD_SEND_R((seq), (line), (id), (argc), NULL, (VALUE)INT2FIX(VM_CALL_FCALL), NULL)
#define ADD_CALL_WITH_BLOCK(seq, line, id, argc, block) \
* rewrite method/block parameter fitting logic to optimize keyword arguments/parameters and a splat argument. [Feature #10440] (Details are described in this ticket) Most of complex part is moved to vm_args.c. Now, ISeq#to_a does not catch up new instruction format. * vm_core.h: change iseq data structures. * introduce rb_call_info_kw_arg_t to represent keyword arguments. * add rb_call_info_t::kw_arg. * rename rb_iseq_t::arg_post_len to rb_iseq_t::arg_post_num. * rename rb_iseq_t::arg_keywords to arg_keyword_num. * rename rb_iseq_t::arg_keyword to rb_iseq_t::arg_keyword_bits. to represent keyword bitmap parameter index. This bitmap parameter shows that which keyword parameters are given or not given (0 for given). It is refered by `checkkeyword' instruction described bellow. * rename rb_iseq_t::arg_keyword_check to rb_iseq_t::arg_keyword_rest to represent keyword rest parameter index. * add rb_iseq_t::arg_keyword_default_values to represent default keyword values. * rename VM_CALL_ARGS_SKIP_SETUP to VM_CALL_ARGS_SIMPLE to represent (ci->flag & (SPLAT|BLOCKARG)) && ci->blockiseq == NULL && ci->kw_arg == NULL. * vm_insnhelper.c, vm_args.c: rewrite with refactoring. * rewrite splat argument code. * rewrite keyword arguments/parameters code. * merge method and block parameter fitting code into one code base. * vm.c, vm_eval.c: catch up these changes. * compile.c (new_callinfo): callinfo requires kw_arg parameter. * compile.c (compile_array_): check the last argument Hash object or not. If Hash object and all keys are Symbol literals, they are compiled to keyword arguments. * insns.def (checkkeyword): add new instruction. This instruction check the availability of corresponding keyword. For example, a method "def foo k1: 'v1'; end" is cimpiled to the following instructions. 0000 checkkeyword 2, 0 # check k1 is given. 0003 branchif 9 # if given, jump to address #9 0005 putstring "v1" 0007 setlocal_OP__WC__0 3 # k1 = 'v1' 0009 trace 8 0011 putnil 0012 trace 16 0014 leave * insns.def (opt_send_simple): removed and add new instruction "opt_send_without_block". * parse.y (new_args_tail_gen): reorder variables. Before this patch, a method "def foo(k1: 1, kr1:, k2: 2, **krest, &b)" has parameter variables "k1, kr1, k2, &b, internal_id, krest", but this patch reorders to "kr1, k1, k2, internal_id, krest, &b". (locate a block variable at last) * parse.y (vtable_pop): added. This function remove latest `n' variables from vtable. * iseq.c: catch up iseq data changes. * proc.c: ditto. * class.c (keyword_error): export as rb_keyword_error(). * common.mk: depend vm_args.c for vm.o. * hash.c (rb_hash_has_key): export. * internal.h: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@48239 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2014-11-02 21:02:55 +03:00
ADD_SEND_R((seq), (line), (id), (argc), (block), (VALUE)INT2FIX(VM_CALL_FCALL), NULL)
* rewrite method/block parameter fitting logic to optimize keyword arguments/parameters and a splat argument. [Feature #10440] (Details are described in this ticket) Most of complex part is moved to vm_args.c. Now, ISeq#to_a does not catch up new instruction format. * vm_core.h: change iseq data structures. * introduce rb_call_info_kw_arg_t to represent keyword arguments. * add rb_call_info_t::kw_arg. * rename rb_iseq_t::arg_post_len to rb_iseq_t::arg_post_num. * rename rb_iseq_t::arg_keywords to arg_keyword_num. * rename rb_iseq_t::arg_keyword to rb_iseq_t::arg_keyword_bits. to represent keyword bitmap parameter index. This bitmap parameter shows that which keyword parameters are given or not given (0 for given). It is refered by `checkkeyword' instruction described bellow. * rename rb_iseq_t::arg_keyword_check to rb_iseq_t::arg_keyword_rest to represent keyword rest parameter index. * add rb_iseq_t::arg_keyword_default_values to represent default keyword values. * rename VM_CALL_ARGS_SKIP_SETUP to VM_CALL_ARGS_SIMPLE to represent (ci->flag & (SPLAT|BLOCKARG)) && ci->blockiseq == NULL && ci->kw_arg == NULL. * vm_insnhelper.c, vm_args.c: rewrite with refactoring. * rewrite splat argument code. * rewrite keyword arguments/parameters code. * merge method and block parameter fitting code into one code base. * vm.c, vm_eval.c: catch up these changes. * compile.c (new_callinfo): callinfo requires kw_arg parameter. * compile.c (compile_array_): check the last argument Hash object or not. If Hash object and all keys are Symbol literals, they are compiled to keyword arguments. * insns.def (checkkeyword): add new instruction. This instruction check the availability of corresponding keyword. For example, a method "def foo k1: 'v1'; end" is cimpiled to the following instructions. 0000 checkkeyword 2, 0 # check k1 is given. 0003 branchif 9 # if given, jump to address #9 0005 putstring "v1" 0007 setlocal_OP__WC__0 3 # k1 = 'v1' 0009 trace 8 0011 putnil 0012 trace 16 0014 leave * insns.def (opt_send_simple): removed and add new instruction "opt_send_without_block". * parse.y (new_args_tail_gen): reorder variables. Before this patch, a method "def foo(k1: 1, kr1:, k2: 2, **krest, &b)" has parameter variables "k1, kr1, k2, &b, internal_id, krest", but this patch reorders to "kr1, k1, k2, internal_id, krest, &b". (locate a block variable at last) * parse.y (vtable_pop): added. This function remove latest `n' variables from vtable. * iseq.c: catch up iseq data changes. * proc.c: ditto. * class.c (keyword_error): export as rb_keyword_error(). * common.mk: depend vm_args.c for vm.o. * hash.c (rb_hash_has_key): export. * internal.h: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@48239 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2014-11-02 21:02:55 +03:00
#define ADD_SEND_R(seq, line, id, argc, block, flag, keywords) \
2015-07-22 01:52:59 +03:00
ADD_ELEM((seq), (LINK_ELEMENT *) new_insn_send(iseq, (line), (id), (VALUE)(argc), (block), (VALUE)(flag), (keywords)))
#define ADD_TRACE(seq, event) \
ADD_ELEM((seq), (LINK_ELEMENT *)new_trace_body(iseq, (event), 0))
#define ADD_TRACE_WITH_DATA(seq, event, data) \
ADD_ELEM((seq), (LINK_ELEMENT *)new_trace_body(iseq, (event), (data)))
#define DECL_BRANCH_BASE(branches, first_line, first_column, last_line, last_column, type) \
do { \
if (ISEQ_COVERAGE(iseq) && \
ISEQ_BRANCH_COVERAGE(iseq) && \
(first_line) > 0) { \
VALUE structure = RARRAY_AREF(ISEQ_BRANCH_COVERAGE(iseq), 0); \
branches = rb_ary_tmp_new(0); \
rb_ary_push(structure, branches); \
rb_ary_push(branches, ID2SYM(rb_intern(type))); \
rb_ary_push(branches, INT2FIX(first_line)); \
rb_ary_push(branches, INT2FIX(first_column)); \
rb_ary_push(branches, INT2FIX(last_line)); \
rb_ary_push(branches, INT2FIX(last_column)); \
} \
} while (0)
#define ADD_TRACE_BRANCH_COVERAGE(seq, first_line, first_column, last_line, last_column, type, branches) \
do { \
if (ISEQ_COVERAGE(iseq) && \
ISEQ_BRANCH_COVERAGE(iseq) && \
(first_line) > 0) { \
VALUE counters = RARRAY_AREF(ISEQ_BRANCH_COVERAGE(iseq), 1); \
long counter_idx = RARRAY_LEN(counters); \
rb_ary_push(counters, INT2FIX(0)); \
rb_ary_push(branches, ID2SYM(rb_intern(type))); \
rb_ary_push(branches, INT2FIX(first_line)); \
rb_ary_push(branches, INT2FIX(first_column)); \
rb_ary_push(branches, INT2FIX(last_line)); \
rb_ary_push(branches, INT2FIX(last_column)); \
rb_ary_push(branches, INT2FIX(counter_idx)); \
ADD_TRACE_WITH_DATA(seq, RUBY_EVENT_COVERAGE_BRANCH, counter_idx); \
ADD_INSN(seq, last_line, nop); \
} \
} while (0)
static void iseq_add_getlocal(rb_iseq_t *iseq, LINK_ANCHOR *const seq, int line, int idx, int level);
static void iseq_add_setlocal(rb_iseq_t *iseq, LINK_ANCHOR *const seq, int line, int idx, int level);
#define ADD_GETLOCAL(seq, line, idx, level) iseq_add_getlocal(iseq, (seq), (line), (idx), (level))
#define ADD_SETLOCAL(seq, line, idx, level) iseq_add_setlocal(iseq, (seq), (line), (idx), (level))
/* add label */
#define ADD_LABEL(seq, label) \
ADD_ELEM((seq), (LINK_ELEMENT *) (label))
#define APPEND_LABEL(seq, before, label) \
APPEND_ELEM((seq), (before), (LINK_ELEMENT *) (label))
#define ADD_ADJUST(seq, line, label) \
ADD_ELEM((seq), (LINK_ELEMENT *) new_adjust_body(iseq, (label), (line)))
#define ADD_ADJUST_RESTORE(seq, label) \
ADD_ELEM((seq), (LINK_ELEMENT *) new_adjust_body(iseq, (label), -1))
#define LABEL_UNREMOVABLE(label) \
((label) ? (LABEL_REF(label), (label)->unremovable=1) : 0)
2015-07-22 01:52:59 +03:00
#define ADD_CATCH_ENTRY(type, ls, le, iseqv, lc) do { \
VALUE _e = rb_ary_new3(5, (type), \
(VALUE)(ls) | 1, (VALUE)(le) | 1, \
(VALUE)(iseqv), (VALUE)(lc) | 1); \
LABEL_UNREMOVABLE(ls); \
LABEL_REF(le); \
LABEL_REF(lc); \
if (NIL_P(ISEQ_COMPILE_DATA(iseq)->catch_table_ary)) \
RB_OBJ_WRITE(iseq, &ISEQ_COMPILE_DATA(iseq)->catch_table_ary, rb_ary_tmp_new(3)); \
rb_ary_push(ISEQ_COMPILE_DATA(iseq)->catch_table_ary, freeze_hide_obj(_e)); \
2015-07-22 01:52:59 +03:00
} while (0)
/* compile node */
#define COMPILE(anchor, desc, node) \
(debug_compile("== " desc "\n", \
iseq_compile_each(iseq, (anchor), (node), 0)))
/* compile node, this node's value will be popped */
#define COMPILE_POPPED(anchor, desc, node) \
(debug_compile("== " desc "\n", \
iseq_compile_each(iseq, (anchor), (node), 1)))
/* compile node, which is popped when 'popped' is true */
#define COMPILE_(anchor, desc, node, popped) \
(debug_compile("== " desc "\n", \
iseq_compile_each(iseq, (anchor), (node), (popped))))
#define COMPILE_RECV(anchor, desc, node) \
(private_recv_p(node) ? \
(ADD_INSN(anchor, nd_line(node), putself), VM_CALL_FCALL) : \
2019-09-19 17:44:37 +03:00
COMPILE(anchor, desc, node->nd_recv) ? 0 : -1)
#define OPERAND_AT(insn, idx) \
(((INSN*)(insn))->operands[(idx)])
#define INSN_OF(insn) \
(((INSN*)(insn))->insn_id)
#define IS_INSN(link) ((link)->type == ISEQ_ELEMENT_INSN)
#define IS_LABEL(link) ((link)->type == ISEQ_ELEMENT_LABEL)
#define IS_ADJUST(link) ((link)->type == ISEQ_ELEMENT_ADJUST)
#define IS_TRACE(link) ((link)->type == ISEQ_ELEMENT_TRACE)
#define IS_INSN_ID(iobj, insn) (INSN_OF(iobj) == BIN(insn))
#define IS_NEXT_INSN_ID(link, insn) \
((link)->next && IS_INSN((link)->next) && IS_INSN_ID((link)->next, insn))
/* error */
#if CPDEBUG > 0
NORETURN(static void append_compile_error(const rb_iseq_t *iseq, int line, const char *fmt, ...));
#endif
static void
append_compile_error(const rb_iseq_t *iseq, int line, const char *fmt, ...)
{
VALUE err_info = ISEQ_COMPILE_DATA(iseq)->err_info;
VALUE file = rb_iseq_path(iseq);
VALUE err = err_info == Qtrue ? Qfalse : err_info;
va_list args;
va_start(args, fmt);
err = rb_syntax_error_append(err, file, line, -1, NULL, fmt, args);
va_end(args);
if (NIL_P(err_info)) {
RB_OBJ_WRITE(iseq, &ISEQ_COMPILE_DATA(iseq)->err_info, err);
rb_set_errinfo(err);
}
else if (!err_info) {
RB_OBJ_WRITE(iseq, &ISEQ_COMPILE_DATA(iseq)->err_info, Qtrue);
}
2019-10-22 20:04:36 +03:00
if (compile_debug) {
if (SPECIAL_CONST_P(err)) err = rb_eSyntaxError;
rb_exc_fatal(err);
}
}
#if 0
static void
compile_bug(rb_iseq_t *iseq, int line, const char *fmt, ...)
{
va_list args;
va_start(args, fmt);
rb_report_bug_valist(rb_iseq_path(iseq), line, fmt, args);
va_end(args);
abort();
}
#endif
#define COMPILE_ERROR append_compile_error
#define ERROR_ARGS_AT(n) iseq, nd_line(n),
#define ERROR_ARGS ERROR_ARGS_AT(node)
#define EXPECT_NODE(prefix, node, ndtype, errval) \
do { \
const NODE *error_node = (node); \
enum node_type error_type = nd_type(error_node); \
if (error_type != (ndtype)) { \
COMPILE_ERROR(ERROR_ARGS_AT(error_node) \
prefix ": " #ndtype " is expected, but %s", \
ruby_node_name(error_type)); \
return errval; \
} \
} while (0)
#define EXPECT_NODE_NONULL(prefix, parent, ndtype, errval) \
do { \
COMPILE_ERROR(ERROR_ARGS_AT(parent) \
prefix ": must be " #ndtype ", but 0"); \
return errval; \
} while (0)
#define UNKNOWN_NODE(prefix, node, errval) \
do { \
const NODE *error_node = (node); \
COMPILE_ERROR(ERROR_ARGS_AT(error_node) prefix ": unknown node (%s)", \
ruby_node_name(nd_type(error_node))); \
return errval; \
} while (0)
#define COMPILE_OK 1
#define COMPILE_NG 0
#define CHECK(sub) if (!(sub)) {BEFORE_RETURN;return COMPILE_NG;}
#define NO_CHECK(sub) (void)(sub)
#define BEFORE_RETURN
/* leave name uninitialized so that compiler warn if INIT_ANCHOR is
* missing */
#define DECL_ANCHOR(name) \
LINK_ANCHOR name[1] = {{{ISEQ_ELEMENT_ANCHOR,},}}
#define INIT_ANCHOR(name) \
(name->last = &name->anchor)
2015-07-22 01:52:59 +03:00
static inline VALUE
freeze_hide_obj(VALUE obj)
{
OBJ_FREEZE(obj);
RBASIC_CLEAR_CLASS(obj);
return obj;
}
#include "optinsn.inc"
#if OPT_INSTRUCTIONS_UNIFICATION
#include "optunifs.inc"
#endif
/* for debug */
#if CPDEBUG < 0
#define ISEQ_ARG iseq,
#define ISEQ_ARG_DECLARE rb_iseq_t *iseq,
#else
#define ISEQ_ARG
#define ISEQ_ARG_DECLARE
#endif
#if CPDEBUG
#define gl_node_level ISEQ_COMPILE_DATA(iseq)->node_level
#endif
static void dump_disasm_list_with_cursor(const LINK_ELEMENT *link, const LINK_ELEMENT *curr, const LABEL *dest);
static void dump_disasm_list(const LINK_ELEMENT *elem);
static int insn_data_length(INSN *iobj);
static int calc_sp_depth(int depth, INSN *iobj);
static INSN *new_insn_body(rb_iseq_t *iseq, int line_no, enum ruby_vminsn_type insn_id, int argc, ...);
static LABEL *new_label_body(rb_iseq_t *iseq, long line);
static ADJUST *new_adjust_body(rb_iseq_t *iseq, LABEL *label, int line);
static TRACE *new_trace_body(rb_iseq_t *iseq, rb_event_flag_t event, long data);
static int iseq_compile_each(rb_iseq_t *iseq, LINK_ANCHOR *anchor, const NODE *n, int);
static int iseq_setup(rb_iseq_t *iseq, LINK_ANCHOR *const anchor);
static int iseq_setup_insn(rb_iseq_t *iseq, LINK_ANCHOR *const anchor);
static int iseq_optimize(rb_iseq_t *iseq, LINK_ANCHOR *const anchor);
static int iseq_insns_unification(rb_iseq_t *iseq, LINK_ANCHOR *const anchor);
static int iseq_set_local_table(rb_iseq_t *iseq, const ID *tbl);
static int iseq_set_exception_local_table(rb_iseq_t *iseq);
static int iseq_set_arguments(rb_iseq_t *iseq, LINK_ANCHOR *const anchor, const NODE *const node);
static int iseq_set_sequence_stackcaching(rb_iseq_t *iseq, LINK_ANCHOR *const anchor);
static int iseq_set_sequence(rb_iseq_t *iseq, LINK_ANCHOR *const anchor);
static int iseq_set_exception_table(rb_iseq_t *iseq);
static int iseq_set_optargs_table(rb_iseq_t *iseq);
static int compile_defined_expr(rb_iseq_t *iseq, LINK_ANCHOR *const ret, const NODE *const node, VALUE needstr);
/*
* To make Array to LinkedList, use link_anchor
*/
static void
verify_list(ISEQ_ARG_DECLARE const char *info, LINK_ANCHOR *const anchor)
{
#if CPDEBUG
int flag = 0;
LINK_ELEMENT *list, *plist;
if (!compile_debug) return;
list = anchor->anchor.next;
plist = &anchor->anchor;
while (list) {
if (plist != list->prev) {
flag += 1;
}
plist = list;
list = list->next;
}
if (anchor->last != plist && anchor->last != 0) {
flag |= 0x70000;
}
if (flag != 0) {
rb_bug("list verify error: %08x (%s)", flag, info);
}
#endif
}
#if CPDEBUG < 0
#define verify_list(info, anchor) verify_list(iseq, (info), (anchor))
#endif
static void
verify_call_cache(rb_iseq_t *iseq)
{
2019-10-21 19:56:44 +03:00
#if CPDEBUG
VALUE *original = rb_iseq_original_iseq(iseq);
size_t i = 0;
while (i < iseq->body->iseq_size) {
VALUE insn = original[i];
const char *types = insn_op_types(insn);
for (int j=0; types[j]; j++) {
if (types[j] == TS_CALLDATA) {
struct rb_call_cache cc;
struct rb_call_data *cd = (struct rb_call_data *)original[i+j+1];
MEMZERO(&cc, cc, 1);
if (memcmp(&cc, &cd->cc, sizeof(cc))) {
rb_bug("call cache not zero for fresh iseq");
}
}
}
i += insn_len(insn);
}
2019-10-21 19:56:44 +03:00
#endif
}
/*
* elem1, elem2 => elem1, elem2, elem
*/
static void
ADD_ELEM(ISEQ_ARG_DECLARE LINK_ANCHOR *const anchor, LINK_ELEMENT *elem)
{
elem->prev = anchor->last;
anchor->last->next = elem;
anchor->last = elem;
verify_list("add", anchor);
}
/*
* elem1, before, elem2 => elem1, before, elem, elem2
*/
static void
APPEND_ELEM(ISEQ_ARG_DECLARE LINK_ANCHOR *const anchor, LINK_ELEMENT *before, LINK_ELEMENT *elem)
{
elem->prev = before;
elem->next = before->next;
elem->next->prev = elem;
before->next = elem;
if (before == anchor->last) anchor->last = elem;
verify_list("add", anchor);
}
#if CPDEBUG < 0
#define ADD_ELEM(anchor, elem) ADD_ELEM(iseq, (anchor), (elem))
#define APPEND_ELEM(anchor, before, elem) APPEND_ELEM(iseq, (anchor), (before), (elem))
#endif
#define ISEQ_LAST_LINE(iseq) (ISEQ_COMPILE_DATA(iseq)->last_line)
static int
validate_label(st_data_t name, st_data_t label, st_data_t arg)
{
rb_iseq_t *iseq = (rb_iseq_t *)arg;
LABEL *lobj = (LABEL *)label;
if (!lobj->link.next) {
do {
COMPILE_ERROR(iseq, lobj->position,
"%"PRIsVALUE": undefined label",
rb_sym2str((VALUE)name));
} while (0);
}
return ST_CONTINUE;
}
static void
validate_labels(rb_iseq_t *iseq, st_table *labels_table)
{
st_foreach(labels_table, validate_label, (st_data_t)iseq);
st_free_table(labels_table);
}
VALUE
rb_iseq_compile_callback(rb_iseq_t *iseq, const struct rb_iseq_new_with_callback_callback_func * ifunc)
{
DECL_ANCHOR(ret);
INIT_ANCHOR(ret);
(*ifunc->func)(iseq, ret, ifunc->data);
ADD_INSN(ret, ISEQ_COMPILE_DATA(iseq)->last_line, leave);
CHECK(iseq_setup_insn(iseq, ret));
return iseq_setup(iseq, ret);
}
VALUE
rb_iseq_compile_node(rb_iseq_t *iseq, const NODE *node)
{
DECL_ANCHOR(ret);
INIT_ANCHOR(ret);
if (imemo_type_p((VALUE)node, imemo_ifunc)) {
rb_raise(rb_eArgError, "unexpected imemo_ifunc");
}
if (node == 0) {
2019-07-15 08:43:39 +03:00
NO_CHECK(COMPILE(ret, "nil", node));
iseq_set_local_table(iseq, 0);
}
/* assume node is T_NODE */
else if (nd_type(node) == NODE_SCOPE) {
/* iseq type of top, method, class, block */
iseq_set_local_table(iseq, node->nd_tbl);
iseq_set_arguments(iseq, ret, node->nd_args);
2015-07-22 01:52:59 +03:00
switch (iseq->body->type) {
case ISEQ_TYPE_BLOCK:
{
LABEL *start = ISEQ_COMPILE_DATA(iseq)->start_label = NEW_LABEL(0);
LABEL *end = ISEQ_COMPILE_DATA(iseq)->end_label = NEW_LABEL(0);
start->rescued = LABEL_RESCUE_BEG;
end->rescued = LABEL_RESCUE_END;
ADD_TRACE(ret, RUBY_EVENT_B_CALL);
ADD_INSN (ret, FIX2INT(iseq->body->location.first_lineno), nop);
ADD_LABEL(ret, start);
CHECK(COMPILE(ret, "block body", node->nd_body));
ADD_LABEL(ret, end);
ADD_TRACE(ret, RUBY_EVENT_B_RETURN);
ISEQ_COMPILE_DATA(iseq)->last_line = iseq->body->location.code_location.end_pos.lineno;
/* wide range catch handler must put at last */
ADD_CATCH_ENTRY(CATCH_TYPE_REDO, start, end, NULL, start);
ADD_CATCH_ENTRY(CATCH_TYPE_NEXT, start, end, NULL, end);
break;
}
case ISEQ_TYPE_CLASS:
{
ADD_TRACE(ret, RUBY_EVENT_CLASS);
CHECK(COMPILE(ret, "scoped node", node->nd_body));
ADD_TRACE(ret, RUBY_EVENT_END);
ISEQ_COMPILE_DATA(iseq)->last_line = nd_line(node);
break;
}
case ISEQ_TYPE_METHOD:
{
ADD_TRACE(ret, RUBY_EVENT_CALL);
CHECK(COMPILE(ret, "scoped node", node->nd_body));
ADD_TRACE(ret, RUBY_EVENT_RETURN);
ISEQ_COMPILE_DATA(iseq)->last_line = nd_line(node);
break;
}
default: {
CHECK(COMPILE(ret, "scoped node", node->nd_body));
break;
}
}
}
else {
const char *m;
#define INVALID_ISEQ_TYPE(type) \
ISEQ_TYPE_##type: m = #type; goto invalid_iseq_type
2015-07-22 01:52:59 +03:00
switch (iseq->body->type) {
case INVALID_ISEQ_TYPE(METHOD);
case INVALID_ISEQ_TYPE(CLASS);
case INVALID_ISEQ_TYPE(BLOCK);
case INVALID_ISEQ_TYPE(EVAL);
case INVALID_ISEQ_TYPE(MAIN);
case INVALID_ISEQ_TYPE(TOP);
#undef INVALID_ISEQ_TYPE /* invalid iseq types end */
case ISEQ_TYPE_RESCUE:
iseq_set_exception_local_table(iseq);
CHECK(COMPILE(ret, "rescue", node));
break;
case ISEQ_TYPE_ENSURE:
iseq_set_exception_local_table(iseq);
CHECK(COMPILE_POPPED(ret, "ensure", node));
break;
case ISEQ_TYPE_PLAIN:
CHECK(COMPILE(ret, "ensure", node));
break;
default:
COMPILE_ERROR(ERROR_ARGS "unknown scope: %d", iseq->body->type);
return COMPILE_NG;
invalid_iseq_type:
COMPILE_ERROR(ERROR_ARGS "compile/ISEQ_TYPE_%s should not be reached", m);
return COMPILE_NG;
}
}
2015-07-22 01:52:59 +03:00
if (iseq->body->type == ISEQ_TYPE_RESCUE || iseq->body->type == ISEQ_TYPE_ENSURE) {
ADD_GETLOCAL(ret, 0, LVAR_ERRINFO, 0);
ADD_INSN1(ret, 0, throw, INT2FIX(0) /* continue throw */ );
}
else {
ADD_INSN(ret, ISEQ_COMPILE_DATA(iseq)->last_line, leave);
}
#if OPT_SUPPORT_JOKE
if (ISEQ_COMPILE_DATA(iseq)->labels_table) {
st_table *labels_table = ISEQ_COMPILE_DATA(iseq)->labels_table;
ISEQ_COMPILE_DATA(iseq)->labels_table = 0;
validate_labels(iseq, labels_table);
}
#endif
CHECK(iseq_setup_insn(iseq, ret));
return iseq_setup(iseq, ret);
}
int
rb_iseq_translate_threaded_code(rb_iseq_t *iseq)
{
#if OPT_DIRECT_THREADED_CODE || OPT_CALL_THREADED_CODE
const void * const *table = rb_vm_get_insns_address_table();
unsigned int i;
VALUE *encoded = (VALUE *)iseq->body->iseq_encoded;
2015-07-22 01:52:59 +03:00
for (i = 0; i < iseq->body->iseq_size; /* */ ) {
int insn = (int)iseq->body->iseq_encoded[i];
int len = insn_len(insn);
encoded[i] = (VALUE)table[insn];
i += len;
}
FL_SET(iseq, ISEQ_TRANSLATED);
#endif
return COMPILE_OK;
}
VALUE *
2015-07-22 01:52:59 +03:00
rb_iseq_original_iseq(const rb_iseq_t *iseq) /* cold path */
{
VALUE *original_code;
if (ISEQ_ORIGINAL_ISEQ(iseq)) return ISEQ_ORIGINAL_ISEQ(iseq);
original_code = ISEQ_ORIGINAL_ISEQ_ALLOC(iseq, iseq->body->iseq_size);
MEMCPY(original_code, iseq->body->iseq_encoded, VALUE, iseq->body->iseq_size);
#if OPT_DIRECT_THREADED_CODE || OPT_CALL_THREADED_CODE
{
unsigned int i;
2015-07-22 01:52:59 +03:00
for (i = 0; i < iseq->body->iseq_size; /* */ ) {
const void *addr = (const void *)original_code[i];
const int insn = rb_vm_insn_addr2insn(addr);
original_code[i] = insn;
i += insn_len(insn);
}
}
#endif
return original_code;
}
/*********************************************/
/* definition of data structure for compiler */
/*********************************************/
/*
* On 32-bit SPARC, GCC by default generates SPARC V7 code that may require
* 8-byte word alignment. On the other hand, Oracle Solaris Studio seems to
* generate SPARCV8PLUS code with unaligned memory access instructions.
* That is why the STRICT_ALIGNMENT is defined only with GCC.
*/
#if defined(__sparc) && SIZEOF_VOIDP == 4 && defined(__GNUC__)
#define STRICT_ALIGNMENT
#endif
#ifdef STRICT_ALIGNMENT
#if defined(HAVE_TRUE_LONG_LONG) && SIZEOF_LONG_LONG > SIZEOF_VALUE
#define ALIGNMENT_SIZE SIZEOF_LONG_LONG
#else
#define ALIGNMENT_SIZE SIZEOF_VALUE
#endif
#define PADDING_SIZE_MAX ((size_t)((ALIGNMENT_SIZE) - 1))
#define ALIGNMENT_SIZE_MASK PADDING_SIZE_MAX
/* Note: ALIGNMENT_SIZE == (2 ** N) is expected. */
#else
#define PADDING_SIZE_MAX 0
#endif /* STRICT_ALIGNMENT */
#ifdef STRICT_ALIGNMENT
/* calculate padding size for aligned memory access */
static size_t
calc_padding(void *ptr, size_t size)
{
size_t mis;
size_t padding = 0;
mis = (size_t)ptr & ALIGNMENT_SIZE_MASK;
if (mis > 0) {
padding = ALIGNMENT_SIZE - mis;
}
/*
* On 32-bit sparc or equivalents, when a single VALUE is requested
* and padding == sizeof(VALUE), it is clear that no padding is needed.
*/
#if ALIGNMENT_SIZE > SIZEOF_VALUE
if (size == sizeof(VALUE) && padding == sizeof(VALUE)) {
padding = 0;
}
#endif
return padding;
}
#endif /* STRICT_ALIGNMENT */
static void *
compile_data_alloc_with_arena(struct iseq_compile_data_storage **arena, size_t size)
{
void *ptr = 0;
struct iseq_compile_data_storage *storage = *arena;
#ifdef STRICT_ALIGNMENT
size_t padding = calc_padding((void *)&storage->buff[storage->pos], size);
#else
const size_t padding = 0; /* expected to be optimized by compiler */
#endif /* STRICT_ALIGNMENT */
if (size >= INT_MAX - padding) rb_memerror();
if (storage->pos + size + padding > storage->size) {
unsigned int alloc_size = storage->size;
while (alloc_size < size + PADDING_SIZE_MAX) {
if (alloc_size >= INT_MAX / 2) rb_memerror();
alloc_size *= 2;
}
storage->next = (void *)ALLOC_N(char, alloc_size +
offsetof(struct iseq_compile_data_storage, buff));
storage = *arena = storage->next;
storage->next = 0;
storage->pos = 0;
storage->size = alloc_size;
#ifdef STRICT_ALIGNMENT
padding = calc_padding((void *)&storage->buff[storage->pos], size);
#endif /* STRICT_ALIGNMENT */
}
#ifdef STRICT_ALIGNMENT
storage->pos += (int)padding;
#endif /* STRICT_ALIGNMENT */
ptr = (void *)&storage->buff[storage->pos];
storage->pos += (int)size;
return ptr;
}
static void *
compile_data_alloc(rb_iseq_t *iseq, size_t size)
{
struct iseq_compile_data_storage ** arena = &ISEQ_COMPILE_DATA(iseq)->node.storage_current;
return compile_data_alloc_with_arena(arena, size);
}
static inline void *
compile_data_alloc2(rb_iseq_t *iseq, size_t x, size_t y)
{
size_t size = rb_size_mul_or_raise(x, y, rb_eRuntimeError);
return compile_data_alloc(iseq, size);
}
static INSN *
compile_data_alloc_insn(rb_iseq_t *iseq)
{
struct iseq_compile_data_storage ** arena = &ISEQ_COMPILE_DATA(iseq)->insn.storage_current;
return (INSN *)compile_data_alloc_with_arena(arena, sizeof(INSN));
}
static LABEL *
compile_data_alloc_label(rb_iseq_t *iseq)
{
return (LABEL *)compile_data_alloc(iseq, sizeof(LABEL));
}
static ADJUST *
compile_data_alloc_adjust(rb_iseq_t *iseq)
{
return (ADJUST *)compile_data_alloc(iseq, sizeof(ADJUST));
}
static TRACE *
compile_data_alloc_trace(rb_iseq_t *iseq)
{
return (TRACE *)compile_data_alloc(iseq, sizeof(TRACE));
}
/*
* elem1, elemX => elem1, elem2, elemX
*/
static void
ELEM_INSERT_NEXT(LINK_ELEMENT *elem1, LINK_ELEMENT *elem2)
{
elem2->next = elem1->next;
elem2->prev = elem1;
elem1->next = elem2;
if (elem2->next) {
elem2->next->prev = elem2;
}
}
/*
* elem1, elemX => elemX, elem2, elem1
*/
static void
ELEM_INSERT_PREV(LINK_ELEMENT *elem1, LINK_ELEMENT *elem2)
{
elem2->prev = elem1->prev;
elem2->next = elem1;
elem1->prev = elem2;
if (elem2->prev) {
elem2->prev->next = elem2;
}
}
/*
* elemX, elem1, elemY => elemX, elem2, elemY
*/
static void
ELEM_REPLACE(LINK_ELEMENT *elem1, LINK_ELEMENT *elem2)
{
elem2->prev = elem1->prev;
elem2->next = elem1->next;
if (elem1->prev) {
elem1->prev->next = elem2;
}
if (elem1->next) {
elem1->next->prev = elem2;
}
}
static void
ELEM_REMOVE(LINK_ELEMENT *elem)
{
elem->prev->next = elem->next;
if (elem->next) {
elem->next->prev = elem->prev;
}
}
static LINK_ELEMENT *
FIRST_ELEMENT(const LINK_ANCHOR *const anchor)
{
return anchor->anchor.next;
}
static LINK_ELEMENT *
LAST_ELEMENT(LINK_ANCHOR *const anchor)
{
return anchor->last;
}
static LINK_ELEMENT *
POP_ELEMENT(ISEQ_ARG_DECLARE LINK_ANCHOR *const anchor)
{
LINK_ELEMENT *elem = anchor->last;
anchor->last = anchor->last->prev;
anchor->last->next = 0;
verify_list("pop", anchor);
return elem;
}
#if CPDEBUG < 0
#define POP_ELEMENT(anchor) POP_ELEMENT(iseq, (anchor))
#endif
static LINK_ELEMENT *
ELEM_FIRST_INSN(LINK_ELEMENT *elem)
{
while (elem) {
switch (elem->type) {
case ISEQ_ELEMENT_INSN:
case ISEQ_ELEMENT_ADJUST:
return elem;
default:
elem = elem->next;
}
}
return NULL;
}
static int
LIST_INSN_SIZE_ONE(const LINK_ANCHOR *const anchor)
{
LINK_ELEMENT *first_insn = ELEM_FIRST_INSN(FIRST_ELEMENT(anchor));
if (first_insn != NULL &&
ELEM_FIRST_INSN(first_insn->next) == NULL) {
return TRUE;
}
else {
return FALSE;
}
}
static int
LIST_INSN_SIZE_ZERO(const LINK_ANCHOR *const anchor)
{
if (ELEM_FIRST_INSN(FIRST_ELEMENT(anchor)) == NULL) {
return TRUE;
}
else {
return FALSE;
}
}
/*
* anc1: e1, e2, e3
* anc2: e4, e5
*#=>
* anc1: e1, e2, e3, e4, e5
* anc2: e4, e5 (broken)
*/
static void
APPEND_LIST(ISEQ_ARG_DECLARE LINK_ANCHOR *const anc1, LINK_ANCHOR *const anc2)
{
if (anc2->anchor.next) {
anc1->last->next = anc2->anchor.next;
anc2->anchor.next->prev = anc1->last;
anc1->last = anc2->last;
}
verify_list("append", anc1);
}
#if CPDEBUG < 0
#define APPEND_LIST(anc1, anc2) APPEND_LIST(iseq, (anc1), (anc2))
#endif
#if CPDEBUG && 0
static void
debug_list(ISEQ_ARG_DECLARE LINK_ANCHOR *const anchor)
{
LINK_ELEMENT *list = FIRST_ELEMENT(anchor);
printf("----\n");
printf("anch: %p, frst: %p, last: %p\n", &anchor->anchor,
anchor->anchor.next, anchor->last);
while (list) {
printf("curr: %p, next: %p, prev: %p, type: %d\n", list, list->next,
list->prev, FIX2INT(list->type));
list = list->next;
}
printf("----\n");
dump_disasm_list(anchor->anchor.next);
verify_list("debug list", anchor);
}
#if CPDEBUG < 0
#define debug_list(anc) debug_list(iseq, (anc))
#endif
#else
#define debug_list(anc) ((void)0)
#endif
static TRACE *
new_trace_body(rb_iseq_t *iseq, rb_event_flag_t event, long data)
{
TRACE *trace = compile_data_alloc_trace(iseq);
trace->link.type = ISEQ_ELEMENT_TRACE;
trace->link.next = NULL;
trace->event = event;
trace->data = data;
return trace;
}
static LABEL *
new_label_body(rb_iseq_t *iseq, long line)
{
LABEL *labelobj = compile_data_alloc_label(iseq);
labelobj->link.type = ISEQ_ELEMENT_LABEL;
labelobj->link.next = 0;
labelobj->label_no = ISEQ_COMPILE_DATA(iseq)->label_no++;
labelobj->sc_state = 0;
labelobj->sp = -1;
labelobj->refcnt = 0;
labelobj->set = 0;
labelobj->rescued = LABEL_RESCUE_NONE;
labelobj->unremovable = 0;
return labelobj;
}
static ADJUST *
new_adjust_body(rb_iseq_t *iseq, LABEL *label, int line)
{
ADJUST *adjust = compile_data_alloc_adjust(iseq);
adjust->link.type = ISEQ_ELEMENT_ADJUST;
adjust->link.next = 0;
adjust->label = label;
adjust->line_no = line;
LABEL_UNREMOVABLE(label);
return adjust;
}
static INSN *
new_insn_core(rb_iseq_t *iseq, int line_no,
int insn_id, int argc, VALUE *argv)
{
INSN *iobj = compile_data_alloc_insn(iseq);
/* printf("insn_id: %d, line: %d\n", insn_id, line_no); */
iobj->link.type = ISEQ_ELEMENT_INSN;
iobj->link.next = 0;
iobj->insn_id = insn_id;
iobj->insn_info.line_no = line_no;
iobj->insn_info.events = 0;
iobj->operands = argv;
iobj->operand_size = argc;
iobj->sc_state = 0;
return iobj;
}
static INSN *
new_insn_body(rb_iseq_t *iseq, int line_no, enum ruby_vminsn_type insn_id, int argc, ...)
{
VALUE *operands = 0;
va_list argv;
if (argc > 0) {
int i;
va_init_list(argv, argc);
operands = compile_data_alloc2(iseq, sizeof(VALUE), argc);
for (i = 0; i < argc; i++) {
VALUE v = va_arg(argv, VALUE);
operands[i] = v;
}
va_end(argv);
}
return new_insn_core(iseq, line_no, insn_id, argc, operands);
}
static struct rb_call_info *
new_callinfo(rb_iseq_t *iseq, ID mid, int argc, unsigned int flag, struct rb_call_info_kw_arg *kw_arg, int has_blockiseq)
{
size_t size = kw_arg != NULL ? sizeof(struct rb_call_info_with_kwarg) : sizeof(struct rb_call_info);
struct rb_call_info *ci = (struct rb_call_info *)compile_data_alloc(iseq, size);
struct rb_call_info_with_kwarg *ci_kw = (struct rb_call_info_with_kwarg *)ci;
* rewrite method/block parameter fitting logic to optimize keyword arguments/parameters and a splat argument. [Feature #10440] (Details are described in this ticket) Most of complex part is moved to vm_args.c. Now, ISeq#to_a does not catch up new instruction format. * vm_core.h: change iseq data structures. * introduce rb_call_info_kw_arg_t to represent keyword arguments. * add rb_call_info_t::kw_arg. * rename rb_iseq_t::arg_post_len to rb_iseq_t::arg_post_num. * rename rb_iseq_t::arg_keywords to arg_keyword_num. * rename rb_iseq_t::arg_keyword to rb_iseq_t::arg_keyword_bits. to represent keyword bitmap parameter index. This bitmap parameter shows that which keyword parameters are given or not given (0 for given). It is refered by `checkkeyword' instruction described bellow. * rename rb_iseq_t::arg_keyword_check to rb_iseq_t::arg_keyword_rest to represent keyword rest parameter index. * add rb_iseq_t::arg_keyword_default_values to represent default keyword values. * rename VM_CALL_ARGS_SKIP_SETUP to VM_CALL_ARGS_SIMPLE to represent (ci->flag & (SPLAT|BLOCKARG)) && ci->blockiseq == NULL && ci->kw_arg == NULL. * vm_insnhelper.c, vm_args.c: rewrite with refactoring. * rewrite splat argument code. * rewrite keyword arguments/parameters code. * merge method and block parameter fitting code into one code base. * vm.c, vm_eval.c: catch up these changes. * compile.c (new_callinfo): callinfo requires kw_arg parameter. * compile.c (compile_array_): check the last argument Hash object or not. If Hash object and all keys are Symbol literals, they are compiled to keyword arguments. * insns.def (checkkeyword): add new instruction. This instruction check the availability of corresponding keyword. For example, a method "def foo k1: 'v1'; end" is cimpiled to the following instructions. 0000 checkkeyword 2, 0 # check k1 is given. 0003 branchif 9 # if given, jump to address #9 0005 putstring "v1" 0007 setlocal_OP__WC__0 3 # k1 = 'v1' 0009 trace 8 0011 putnil 0012 trace 16 0014 leave * insns.def (opt_send_simple): removed and add new instruction "opt_send_without_block". * parse.y (new_args_tail_gen): reorder variables. Before this patch, a method "def foo(k1: 1, kr1:, k2: 2, **krest, &b)" has parameter variables "k1, kr1, k2, &b, internal_id, krest", but this patch reorders to "kr1, k1, k2, internal_id, krest, &b". (locate a block variable at last) * parse.y (vtable_pop): added. This function remove latest `n' variables from vtable. * iseq.c: catch up iseq data changes. * proc.c: ditto. * class.c (keyword_error): export as rb_keyword_error(). * common.mk: depend vm_args.c for vm.o. * hash.c (rb_hash_has_key): export. * internal.h: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@48239 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2014-11-02 21:02:55 +03:00
ci->mid = mid;
ci->flag = flag;
ci->orig_argc = argc;
* rewrite method/block parameter fitting logic to optimize keyword arguments/parameters and a splat argument. [Feature #10440] (Details are described in this ticket) Most of complex part is moved to vm_args.c. Now, ISeq#to_a does not catch up new instruction format. * vm_core.h: change iseq data structures. * introduce rb_call_info_kw_arg_t to represent keyword arguments. * add rb_call_info_t::kw_arg. * rename rb_iseq_t::arg_post_len to rb_iseq_t::arg_post_num. * rename rb_iseq_t::arg_keywords to arg_keyword_num. * rename rb_iseq_t::arg_keyword to rb_iseq_t::arg_keyword_bits. to represent keyword bitmap parameter index. This bitmap parameter shows that which keyword parameters are given or not given (0 for given). It is refered by `checkkeyword' instruction described bellow. * rename rb_iseq_t::arg_keyword_check to rb_iseq_t::arg_keyword_rest to represent keyword rest parameter index. * add rb_iseq_t::arg_keyword_default_values to represent default keyword values. * rename VM_CALL_ARGS_SKIP_SETUP to VM_CALL_ARGS_SIMPLE to represent (ci->flag & (SPLAT|BLOCKARG)) && ci->blockiseq == NULL && ci->kw_arg == NULL. * vm_insnhelper.c, vm_args.c: rewrite with refactoring. * rewrite splat argument code. * rewrite keyword arguments/parameters code. * merge method and block parameter fitting code into one code base. * vm.c, vm_eval.c: catch up these changes. * compile.c (new_callinfo): callinfo requires kw_arg parameter. * compile.c (compile_array_): check the last argument Hash object or not. If Hash object and all keys are Symbol literals, they are compiled to keyword arguments. * insns.def (checkkeyword): add new instruction. This instruction check the availability of corresponding keyword. For example, a method "def foo k1: 'v1'; end" is cimpiled to the following instructions. 0000 checkkeyword 2, 0 # check k1 is given. 0003 branchif 9 # if given, jump to address #9 0005 putstring "v1" 0007 setlocal_OP__WC__0 3 # k1 = 'v1' 0009 trace 8 0011 putnil 0012 trace 16 0014 leave * insns.def (opt_send_simple): removed and add new instruction "opt_send_without_block". * parse.y (new_args_tail_gen): reorder variables. Before this patch, a method "def foo(k1: 1, kr1:, k2: 2, **krest, &b)" has parameter variables "k1, kr1, k2, &b, internal_id, krest", but this patch reorders to "kr1, k1, k2, internal_id, krest, &b". (locate a block variable at last) * parse.y (vtable_pop): added. This function remove latest `n' variables from vtable. * iseq.c: catch up iseq data changes. * proc.c: ditto. * class.c (keyword_error): export as rb_keyword_error(). * common.mk: depend vm_args.c for vm.o. * hash.c (rb_hash_has_key): export. * internal.h: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@48239 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2014-11-02 21:02:55 +03:00
if (kw_arg) {
ci->flag |= VM_CALL_KWARG;
ci_kw->kw_arg = kw_arg;
* rewrite method/block parameter fitting logic to optimize keyword arguments/parameters and a splat argument. [Feature #10440] (Details are described in this ticket) Most of complex part is moved to vm_args.c. Now, ISeq#to_a does not catch up new instruction format. * vm_core.h: change iseq data structures. * introduce rb_call_info_kw_arg_t to represent keyword arguments. * add rb_call_info_t::kw_arg. * rename rb_iseq_t::arg_post_len to rb_iseq_t::arg_post_num. * rename rb_iseq_t::arg_keywords to arg_keyword_num. * rename rb_iseq_t::arg_keyword to rb_iseq_t::arg_keyword_bits. to represent keyword bitmap parameter index. This bitmap parameter shows that which keyword parameters are given or not given (0 for given). It is refered by `checkkeyword' instruction described bellow. * rename rb_iseq_t::arg_keyword_check to rb_iseq_t::arg_keyword_rest to represent keyword rest parameter index. * add rb_iseq_t::arg_keyword_default_values to represent default keyword values. * rename VM_CALL_ARGS_SKIP_SETUP to VM_CALL_ARGS_SIMPLE to represent (ci->flag & (SPLAT|BLOCKARG)) && ci->blockiseq == NULL && ci->kw_arg == NULL. * vm_insnhelper.c, vm_args.c: rewrite with refactoring. * rewrite splat argument code. * rewrite keyword arguments/parameters code. * merge method and block parameter fitting code into one code base. * vm.c, vm_eval.c: catch up these changes. * compile.c (new_callinfo): callinfo requires kw_arg parameter. * compile.c (compile_array_): check the last argument Hash object or not. If Hash object and all keys are Symbol literals, they are compiled to keyword arguments. * insns.def (checkkeyword): add new instruction. This instruction check the availability of corresponding keyword. For example, a method "def foo k1: 'v1'; end" is cimpiled to the following instructions. 0000 checkkeyword 2, 0 # check k1 is given. 0003 branchif 9 # if given, jump to address #9 0005 putstring "v1" 0007 setlocal_OP__WC__0 3 # k1 = 'v1' 0009 trace 8 0011 putnil 0012 trace 16 0014 leave * insns.def (opt_send_simple): removed and add new instruction "opt_send_without_block". * parse.y (new_args_tail_gen): reorder variables. Before this patch, a method "def foo(k1: 1, kr1:, k2: 2, **krest, &b)" has parameter variables "k1, kr1, k2, &b, internal_id, krest", but this patch reorders to "kr1, k1, k2, internal_id, krest, &b". (locate a block variable at last) * parse.y (vtable_pop): added. This function remove latest `n' variables from vtable. * iseq.c: catch up iseq data changes. * proc.c: ditto. * class.c (keyword_error): export as rb_keyword_error(). * common.mk: depend vm_args.c for vm.o. * hash.c (rb_hash_has_key): export. * internal.h: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@48239 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2014-11-02 21:02:55 +03:00
ci->orig_argc += kw_arg->keyword_len;
iseq->body->ci_kw_size++;
}
else {
iseq->body->ci_size++;
* rewrite method/block parameter fitting logic to optimize keyword arguments/parameters and a splat argument. [Feature #10440] (Details are described in this ticket) Most of complex part is moved to vm_args.c. Now, ISeq#to_a does not catch up new instruction format. * vm_core.h: change iseq data structures. * introduce rb_call_info_kw_arg_t to represent keyword arguments. * add rb_call_info_t::kw_arg. * rename rb_iseq_t::arg_post_len to rb_iseq_t::arg_post_num. * rename rb_iseq_t::arg_keywords to arg_keyword_num. * rename rb_iseq_t::arg_keyword to rb_iseq_t::arg_keyword_bits. to represent keyword bitmap parameter index. This bitmap parameter shows that which keyword parameters are given or not given (0 for given). It is refered by `checkkeyword' instruction described bellow. * rename rb_iseq_t::arg_keyword_check to rb_iseq_t::arg_keyword_rest to represent keyword rest parameter index. * add rb_iseq_t::arg_keyword_default_values to represent default keyword values. * rename VM_CALL_ARGS_SKIP_SETUP to VM_CALL_ARGS_SIMPLE to represent (ci->flag & (SPLAT|BLOCKARG)) && ci->blockiseq == NULL && ci->kw_arg == NULL. * vm_insnhelper.c, vm_args.c: rewrite with refactoring. * rewrite splat argument code. * rewrite keyword arguments/parameters code. * merge method and block parameter fitting code into one code base. * vm.c, vm_eval.c: catch up these changes. * compile.c (new_callinfo): callinfo requires kw_arg parameter. * compile.c (compile_array_): check the last argument Hash object or not. If Hash object and all keys are Symbol literals, they are compiled to keyword arguments. * insns.def (checkkeyword): add new instruction. This instruction check the availability of corresponding keyword. For example, a method "def foo k1: 'v1'; end" is cimpiled to the following instructions. 0000 checkkeyword 2, 0 # check k1 is given. 0003 branchif 9 # if given, jump to address #9 0005 putstring "v1" 0007 setlocal_OP__WC__0 3 # k1 = 'v1' 0009 trace 8 0011 putnil 0012 trace 16 0014 leave * insns.def (opt_send_simple): removed and add new instruction "opt_send_without_block". * parse.y (new_args_tail_gen): reorder variables. Before this patch, a method "def foo(k1: 1, kr1:, k2: 2, **krest, &b)" has parameter variables "k1, kr1, k2, &b, internal_id, krest", but this patch reorders to "kr1, k1, k2, internal_id, krest, &b". (locate a block variable at last) * parse.y (vtable_pop): added. This function remove latest `n' variables from vtable. * iseq.c: catch up iseq data changes. * proc.c: ditto. * class.c (keyword_error): export as rb_keyword_error(). * common.mk: depend vm_args.c for vm.o. * hash.c (rb_hash_has_key): export. * internal.h: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@48239 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2014-11-02 21:02:55 +03:00
}
if (!(ci->flag & (VM_CALL_ARGS_SPLAT | VM_CALL_ARGS_BLOCKARG | VM_CALL_KW_SPLAT)) &&
kw_arg == NULL && !has_blockiseq) {
* rewrite method/block parameter fitting logic to optimize keyword arguments/parameters and a splat argument. [Feature #10440] (Details are described in this ticket) Most of complex part is moved to vm_args.c. Now, ISeq#to_a does not catch up new instruction format. * vm_core.h: change iseq data structures. * introduce rb_call_info_kw_arg_t to represent keyword arguments. * add rb_call_info_t::kw_arg. * rename rb_iseq_t::arg_post_len to rb_iseq_t::arg_post_num. * rename rb_iseq_t::arg_keywords to arg_keyword_num. * rename rb_iseq_t::arg_keyword to rb_iseq_t::arg_keyword_bits. to represent keyword bitmap parameter index. This bitmap parameter shows that which keyword parameters are given or not given (0 for given). It is refered by `checkkeyword' instruction described bellow. * rename rb_iseq_t::arg_keyword_check to rb_iseq_t::arg_keyword_rest to represent keyword rest parameter index. * add rb_iseq_t::arg_keyword_default_values to represent default keyword values. * rename VM_CALL_ARGS_SKIP_SETUP to VM_CALL_ARGS_SIMPLE to represent (ci->flag & (SPLAT|BLOCKARG)) && ci->blockiseq == NULL && ci->kw_arg == NULL. * vm_insnhelper.c, vm_args.c: rewrite with refactoring. * rewrite splat argument code. * rewrite keyword arguments/parameters code. * merge method and block parameter fitting code into one code base. * vm.c, vm_eval.c: catch up these changes. * compile.c (new_callinfo): callinfo requires kw_arg parameter. * compile.c (compile_array_): check the last argument Hash object or not. If Hash object and all keys are Symbol literals, they are compiled to keyword arguments. * insns.def (checkkeyword): add new instruction. This instruction check the availability of corresponding keyword. For example, a method "def foo k1: 'v1'; end" is cimpiled to the following instructions. 0000 checkkeyword 2, 0 # check k1 is given. 0003 branchif 9 # if given, jump to address #9 0005 putstring "v1" 0007 setlocal_OP__WC__0 3 # k1 = 'v1' 0009 trace 8 0011 putnil 0012 trace 16 0014 leave * insns.def (opt_send_simple): removed and add new instruction "opt_send_without_block". * parse.y (new_args_tail_gen): reorder variables. Before this patch, a method "def foo(k1: 1, kr1:, k2: 2, **krest, &b)" has parameter variables "k1, kr1, k2, &b, internal_id, krest", but this patch reorders to "kr1, k1, k2, internal_id, krest, &b". (locate a block variable at last) * parse.y (vtable_pop): added. This function remove latest `n' variables from vtable. * iseq.c: catch up iseq data changes. * proc.c: ditto. * class.c (keyword_error): export as rb_keyword_error(). * common.mk: depend vm_args.c for vm.o. * hash.c (rb_hash_has_key): export. * internal.h: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@48239 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2014-11-02 21:02:55 +03:00
ci->flag |= VM_CALL_ARGS_SIMPLE;
}
return ci;
}
static INSN *
new_insn_send(rb_iseq_t *iseq, int line_no, ID id, VALUE argc, const rb_iseq_t *blockiseq, VALUE flag, struct rb_call_info_kw_arg *keywords)
{
VALUE *operands = compile_data_alloc2(iseq, sizeof(VALUE), 2);
operands[0] = (VALUE)new_callinfo(iseq, id, FIX2INT(argc), FIX2INT(flag), keywords, blockiseq != NULL);
operands[1] = (VALUE)blockiseq;
return new_insn_core(iseq, line_no, BIN(send), 2, operands);
}
2015-07-22 01:52:59 +03:00
static rb_iseq_t *
new_child_iseq(rb_iseq_t *iseq, const NODE *const node,
2015-07-22 01:52:59 +03:00
VALUE name, const rb_iseq_t *parent, enum iseq_type type, int line_no)
{
2015-07-22 01:52:59 +03:00
rb_iseq_t *ret_iseq;
rb_ast_body_t ast;
ast.root = node;
ast.compile_option = 0;
ast.line_count = -1;
debugs("[new_child_iseq]> ---------------------------------------\n");
ret_iseq = rb_iseq_new_with_opt(&ast, name,
rb_iseq_path(iseq), rb_iseq_realpath(iseq),
INT2FIX(line_no), parent, type, ISEQ_COMPILE_DATA(iseq)->option);
debugs("[new_child_iseq]< ---------------------------------------\n");
2015-07-22 01:52:59 +03:00
return ret_iseq;
}
static rb_iseq_t *
new_child_iseq_with_callback(rb_iseq_t *iseq, const struct rb_iseq_new_with_callback_callback_func *ifunc,
VALUE name, const rb_iseq_t *parent, enum iseq_type type, int line_no)
{
rb_iseq_t *ret_iseq;
debugs("[new_child_iseq_with_callback]> ---------------------------------------\n");
ret_iseq = rb_iseq_new_with_callback(ifunc, name,
rb_iseq_path(iseq), rb_iseq_realpath(iseq),
INT2FIX(line_no), parent, type, ISEQ_COMPILE_DATA(iseq)->option);
debugs("[new_child_iseq_with_callback]< ---------------------------------------\n");
return ret_iseq;
}
static void
set_catch_except_p(struct rb_iseq_constant_body *body)
{
body->catch_except_p = TRUE;
if (body->parent_iseq != NULL) {
set_catch_except_p(body->parent_iseq->body);
}
}
/* Set body->catch_except_p to TRUE if the ISeq may catch an exception. If it is FALSE,
JIT-ed code may be optimized. If we are extremely conservative, we should set TRUE
if catch table exists. But we want to optimize while loop, which always has catch
table entries for break/next/redo.
So this function sets TRUE for limited ISeqs with break/next/redo catch table entries
whose child ISeq would really raise an exception. */
static void
update_catch_except_flags(struct rb_iseq_constant_body *body)
{
unsigned int pos;
size_t i;
int insn;
const struct iseq_catch_table *ct = body->catch_table;
/* This assumes that a block has parent_iseq which may catch an exception from the block, and that
BREAK/NEXT/REDO catch table entries are used only when `throw` insn is used in the block. */
pos = 0;
while (pos < body->iseq_size) {
#if OPT_DIRECT_THREADED_CODE || OPT_CALL_THREADED_CODE
insn = rb_vm_insn_addr2insn((void *)body->iseq_encoded[pos]);
#else
insn = (int)body->iseq_encoded[pos];
#endif
if (insn == BIN(throw)) {
set_catch_except_p(body);
break;
}
pos += insn_len(insn);
}
if (ct == NULL)
return;
for (i = 0; i < ct->size; i++) {
const struct iseq_catch_table_entry *entry =
UNALIGNED_MEMBER_PTR(ct, entries[i]);
if (entry->type != CATCH_TYPE_BREAK
&& entry->type != CATCH_TYPE_NEXT
&& entry->type != CATCH_TYPE_REDO) {
body->catch_except_p = TRUE;
break;
}
}
}
static void
iseq_insert_nop_between_end_and_cont(rb_iseq_t *iseq)
{
VALUE catch_table_ary = ISEQ_COMPILE_DATA(iseq)->catch_table_ary;
if (NIL_P(catch_table_ary)) return;
unsigned int i, tlen = (unsigned int)RARRAY_LEN(catch_table_ary);
const VALUE *tptr = RARRAY_CONST_PTR_TRANSIENT(catch_table_ary);
for (i = 0; i < tlen; i++) {
const VALUE *ptr = RARRAY_CONST_PTR_TRANSIENT(tptr[i]);
LINK_ELEMENT *end = (LINK_ELEMENT *)(ptr[2] & ~1);
LINK_ELEMENT *cont = (LINK_ELEMENT *)(ptr[4] & ~1);
LINK_ELEMENT *e;
for (e = end; e && (IS_LABEL(e) || IS_TRACE(e)); e = e->next) {
if (e == cont) {
INSN *nop = new_insn_core(iseq, 0, BIN(nop), 0, 0);
ELEM_INSERT_NEXT(end, &nop->link);
break;
}
}
}
}
static int
iseq_setup_insn(rb_iseq_t *iseq, LINK_ANCHOR *const anchor)
{
if (RTEST(ISEQ_COMPILE_DATA(iseq)->err_info))
return COMPILE_NG;
/* debugs("[compile step 2] (iseq_array_to_linkedlist)\n"); */
if (compile_debug > 5)
dump_disasm_list(FIRST_ELEMENT(anchor));
debugs("[compile step 3.1 (iseq_optimize)]\n");
iseq_optimize(iseq, anchor);
if (compile_debug > 5)
dump_disasm_list(FIRST_ELEMENT(anchor));
if (ISEQ_COMPILE_DATA(iseq)->option->instructions_unification) {
debugs("[compile step 3.2 (iseq_insns_unification)]\n");
iseq_insns_unification(iseq, anchor);
if (compile_debug > 5)
dump_disasm_list(FIRST_ELEMENT(anchor));
}
if (ISEQ_COMPILE_DATA(iseq)->option->stack_caching) {
debugs("[compile step 3.3 (iseq_set_sequence_stackcaching)]\n");
iseq_set_sequence_stackcaching(iseq, anchor);
if (compile_debug > 5)
dump_disasm_list(FIRST_ELEMENT(anchor));
}
debugs("[compile step 3.4 (iseq_insert_nop_between_end_and_cont)]\n");
iseq_insert_nop_between_end_and_cont(iseq);
return COMPILE_OK;
}
static int
iseq_setup(rb_iseq_t *iseq, LINK_ANCHOR *const anchor)
{
if (RTEST(ISEQ_COMPILE_DATA(iseq)->err_info))
return COMPILE_NG;
debugs("[compile step 4.1 (iseq_set_sequence)]\n");
if (!iseq_set_sequence(iseq, anchor)) return COMPILE_NG;
if (compile_debug > 5)
dump_disasm_list(FIRST_ELEMENT(anchor));
debugs("[compile step 4.2 (iseq_set_exception_table)]\n");
if (!iseq_set_exception_table(iseq)) return COMPILE_NG;
debugs("[compile step 4.3 (set_optargs_table)] \n");
if (!iseq_set_optargs_table(iseq)) return COMPILE_NG;
debugs("[compile step 5 (iseq_translate_threaded_code)] \n");
if (!rb_iseq_translate_threaded_code(iseq)) return COMPILE_NG;
update_catch_except_flags(iseq->body);
if (compile_debug > 1) {
2015-07-22 01:52:59 +03:00
VALUE str = rb_iseq_disasm(iseq);
printf("%s\n", StringValueCStr(str));
}
verify_call_cache(iseq);
debugs("[compile step: finish]\n");
return COMPILE_OK;
}
static int
iseq_set_exception_local_table(rb_iseq_t *iseq)
{
iseq->body->local_table_size = numberof(rb_iseq_shared_exc_local_tbl);
iseq->body->local_table = rb_iseq_shared_exc_local_tbl;
return COMPILE_OK;
}
static int
get_lvar_level(const rb_iseq_t *iseq)
{
int lev = 0;
2015-07-22 01:52:59 +03:00
while (iseq != iseq->body->local_iseq) {
lev++;
2015-07-22 01:52:59 +03:00
iseq = iseq->body->parent_iseq;
}
return lev;
}
static int
get_dyna_var_idx_at_raw(const rb_iseq_t *iseq, ID id)
{
unsigned int i;
2015-07-22 01:52:59 +03:00
for (i = 0; i < iseq->body->local_table_size; i++) {
if (iseq->body->local_table[i] == id) {
return (int)i;
}
}
return -1;
}
static int
get_local_var_idx(const rb_iseq_t *iseq, ID id)
{
2015-07-22 01:52:59 +03:00
int idx = get_dyna_var_idx_at_raw(iseq->body->local_iseq, id);
if (idx < 0) {
COMPILE_ERROR(iseq, ISEQ_LAST_LINE(iseq),
"get_local_var_idx: %d", idx);
}
return idx;
}
static int
get_dyna_var_idx(const rb_iseq_t *iseq, ID id, int *level, int *ls)
{
int lv = 0, idx = -1;
const rb_iseq_t *const topmost_iseq = iseq;
while (iseq) {
idx = get_dyna_var_idx_at_raw(iseq, id);
if (idx >= 0) {
break;
}
2015-07-22 01:52:59 +03:00
iseq = iseq->body->parent_iseq;
lv++;
}
if (idx < 0) {
COMPILE_ERROR(topmost_iseq, ISEQ_LAST_LINE(topmost_iseq),
"get_dyna_var_idx: -1");
}
*level = lv;
*ls = iseq->body->local_table_size;
return idx;
}
static int
iseq_local_block_param_p(const rb_iseq_t *iseq, unsigned int idx, unsigned int level)
{
const struct rb_iseq_constant_body *body;
while (level > 0) {
iseq = iseq->body->parent_iseq;
level--;
}
body = iseq->body;
if (body->local_iseq == iseq && /* local variables */
body->param.flags.has_block &&
body->local_table_size - body->param.block_start == idx) {
return TRUE;
}
else {
return FALSE;
}
}
static int
iseq_block_param_id_p(const rb_iseq_t *iseq, ID id, int *pidx, int *plevel)
{
int level, ls;
int idx = get_dyna_var_idx(iseq, id, &level, &ls);
if (iseq_local_block_param_p(iseq, ls - idx, level)) {
*pidx = ls - idx;
*plevel = level;
return TRUE;
}
else {
return FALSE;
}
}
static void
iseq_add_getlocal(rb_iseq_t *iseq, LINK_ANCHOR *const seq, int line, int idx, int level)
{
if (iseq_local_block_param_p(iseq, idx, level)) {
ADD_INSN2(seq, line, getblockparam, INT2FIX((idx) + VM_ENV_DATA_SIZE - 1), INT2FIX(level));
}
else {
ADD_INSN2(seq, line, getlocal, INT2FIX((idx) + VM_ENV_DATA_SIZE - 1), INT2FIX(level));
}
}
static void
iseq_add_setlocal(rb_iseq_t *iseq, LINK_ANCHOR *const seq, int line, int idx, int level)
{
if (iseq_local_block_param_p(iseq, idx, level)) {
ADD_INSN2(seq, line, setblockparam, INT2FIX((idx) + VM_ENV_DATA_SIZE - 1), INT2FIX(level));
}
else {
ADD_INSN2(seq, line, setlocal, INT2FIX((idx) + VM_ENV_DATA_SIZE - 1), INT2FIX(level));
}
}
static void
iseq_calc_param_size(rb_iseq_t *iseq)
{
struct rb_iseq_constant_body *const body = iseq->body;
if (body->param.flags.has_opt ||
body->param.flags.has_post ||
body->param.flags.has_rest ||
body->param.flags.has_block ||
body->param.flags.has_kw ||
body->param.flags.has_kwrest) {
if (body->param.flags.has_block) {
body->param.size = body->param.block_start + 1;
}
else if (body->param.flags.has_kwrest) {
body->param.size = body->param.keyword->rest_start + 1;
}
else if (body->param.flags.has_kw) {
body->param.size = body->param.keyword->bits_start + 1;
}
else if (body->param.flags.has_post) {
body->param.size = body->param.post_start + body->param.post_num;
}
else if (body->param.flags.has_rest) {
body->param.size = body->param.rest_start + 1;
}
else if (body->param.flags.has_opt) {
body->param.size = body->param.lead_num + body->param.opt_num;
}
else {
UNREACHABLE;
}
}
else {
body->param.size = body->param.lead_num;
}
}
static int
iseq_set_arguments_keywords(rb_iseq_t *iseq, LINK_ANCHOR *const optargs,
const struct rb_args_info *args, int arg_size)
{
const NODE *node = args->kw_args;
struct rb_iseq_constant_body *const body = iseq->body;
struct rb_iseq_param_keyword *keyword;
const VALUE default_values = rb_ary_tmp_new(1);
const VALUE complex_mark = rb_str_tmp_new(0);
int kw = 0, rkw = 0, di = 0, i;
body->param.flags.has_kw = TRUE;
body->param.keyword = keyword = ZALLOC_N(struct rb_iseq_param_keyword, 1);
while (node) {
kw++;
node = node->nd_next;
}
arg_size += kw;
keyword->bits_start = arg_size++;
node = args->kw_args;
while (node) {
const NODE *val_node = node->nd_body->nd_value;
VALUE dv;
2019-06-04 17:17:38 +03:00
if (val_node == NODE_SPECIAL_REQUIRED_KEYWORD) {
++rkw;
}
else {
switch (nd_type(val_node)) {
case NODE_LIT:
dv = val_node->nd_lit;
break;
case NODE_NIL:
dv = Qnil;
break;
case NODE_TRUE:
dv = Qtrue;
break;
case NODE_FALSE:
dv = Qfalse;
break;
default:
2019-07-15 08:43:39 +03:00
NO_CHECK(COMPILE_POPPED(optargs, "kwarg", node)); /* nd_type(node) == NODE_KW_ARG */
dv = complex_mark;
}
keyword->num = ++di;
rb_ary_push(default_values, dv);
}
node = node->nd_next;
}
keyword->num = kw;
if (args->kw_rest_arg->nd_vid != 0) {
keyword->rest_start = arg_size++;
body->param.flags.has_kwrest = TRUE;
}
keyword->required_num = rkw;
keyword->table = &body->local_table[keyword->bits_start - keyword->num];
{
VALUE *dvs = ALLOC_N(VALUE, RARRAY_LEN(default_values));
for (i = 0; i < RARRAY_LEN(default_values); i++) {
VALUE dv = RARRAY_AREF(default_values, i);
if (dv == complex_mark) dv = Qundef;
if (!SPECIAL_CONST_P(dv)) {
RB_OBJ_WRITTEN(iseq, Qundef, dv);
}
dvs[i] = dv;
}
keyword->default_values = dvs;
}
return arg_size;
}
static int
iseq_set_arguments(rb_iseq_t *iseq, LINK_ANCHOR *const optargs, const NODE *const node_args)
{
debugs("iseq_set_arguments: %s\n", node_args ? "" : "0");
if (node_args) {
struct rb_iseq_constant_body *const body = iseq->body;
struct rb_args_info *args = node_args->nd_ainfo;
ID rest_id = 0;
int last_comma = 0;
ID block_id = 0;
int arg_size;
EXPECT_NODE("iseq_set_arguments", node_args, NODE_ARGS, COMPILE_NG);
body->param.lead_num = arg_size = (int)args->pre_args_num;
if (body->param.lead_num > 0) body->param.flags.has_lead = TRUE;
debugs(" - argc: %d\n", body->param.lead_num);
rest_id = args->rest_arg;
if (rest_id == NODE_SPECIAL_EXCESSIVE_COMMA) {
last_comma = 1;
rest_id = 0;
}
block_id = args->block_arg;
if (args->opt_args) {
const NODE *node = args->opt_args;
LABEL *label;
VALUE labels = rb_ary_tmp_new(1);
VALUE *opt_table;
int i = 0, j;
while (node) {
label = NEW_LABEL(nd_line(node));
rb_ary_push(labels, (VALUE)label | 1);
ADD_LABEL(optargs, label);
2019-07-17 17:42:21 +03:00
NO_CHECK(COMPILE_POPPED(optargs, "optarg", node->nd_body));
node = node->nd_next;
i += 1;
}
/* last label */
label = NEW_LABEL(nd_line(node_args));
rb_ary_push(labels, (VALUE)label | 1);
ADD_LABEL(optargs, label);
opt_table = ALLOC_N(VALUE, i+1);
MEMCPY(opt_table, RARRAY_CONST_PTR_TRANSIENT(labels), VALUE, i+1);
for (j = 0; j < i+1; j++) {
opt_table[j] &= ~1;
}
rb_ary_clear(labels);
body->param.flags.has_opt = TRUE;
body->param.opt_num = i;
body->param.opt_table = opt_table;
arg_size += i;
}
if (rest_id) {
body->param.rest_start = arg_size++;
body->param.flags.has_rest = TRUE;
assert(body->param.rest_start != -1);
}
if (args->first_post_arg) {
body->param.post_start = arg_size;
body->param.post_num = args->post_args_num;
body->param.flags.has_post = TRUE;
arg_size += args->post_args_num;
if (body->param.flags.has_rest) { /* TODO: why that? */
body->param.post_start = body->param.rest_start + 1;
}
}
if (args->kw_args) {
arg_size = iseq_set_arguments_keywords(iseq, optargs, args, arg_size);
}
else if (args->kw_rest_arg) {
struct rb_iseq_param_keyword *keyword = ZALLOC_N(struct rb_iseq_param_keyword, 1);
keyword->rest_start = arg_size++;
body->param.keyword = keyword;
body->param.flags.has_kwrest = TRUE;
}
else if (args->no_kwarg) {
body->param.flags.accepts_no_kwarg = TRUE;
}
if (block_id) {
body->param.block_start = arg_size++;
body->param.flags.has_block = TRUE;
}
iseq_calc_param_size(iseq);
body->param.size = arg_size;
if (args->pre_init) { /* m_init */
2019-07-16 01:19:52 +03:00
NO_CHECK(COMPILE_POPPED(optargs, "init arguments (m)", args->pre_init));
}
if (args->post_init) { /* p_init */
2019-07-15 08:43:39 +03:00
NO_CHECK(COMPILE_POPPED(optargs, "init arguments (p)", args->post_init));
}
if (body->type == ISEQ_TYPE_BLOCK) {
if (body->param.flags.has_opt == FALSE &&
body->param.flags.has_post == FALSE &&
body->param.flags.has_rest == FALSE &&
body->param.flags.has_kw == FALSE &&
body->param.flags.has_kwrest == FALSE) {
* rewrite method/block parameter fitting logic to optimize keyword arguments/parameters and a splat argument. [Feature #10440] (Details are described in this ticket) Most of complex part is moved to vm_args.c. Now, ISeq#to_a does not catch up new instruction format. * vm_core.h: change iseq data structures. * introduce rb_call_info_kw_arg_t to represent keyword arguments. * add rb_call_info_t::kw_arg. * rename rb_iseq_t::arg_post_len to rb_iseq_t::arg_post_num. * rename rb_iseq_t::arg_keywords to arg_keyword_num. * rename rb_iseq_t::arg_keyword to rb_iseq_t::arg_keyword_bits. to represent keyword bitmap parameter index. This bitmap parameter shows that which keyword parameters are given or not given (0 for given). It is refered by `checkkeyword' instruction described bellow. * rename rb_iseq_t::arg_keyword_check to rb_iseq_t::arg_keyword_rest to represent keyword rest parameter index. * add rb_iseq_t::arg_keyword_default_values to represent default keyword values. * rename VM_CALL_ARGS_SKIP_SETUP to VM_CALL_ARGS_SIMPLE to represent (ci->flag & (SPLAT|BLOCKARG)) && ci->blockiseq == NULL && ci->kw_arg == NULL. * vm_insnhelper.c, vm_args.c: rewrite with refactoring. * rewrite splat argument code. * rewrite keyword arguments/parameters code. * merge method and block parameter fitting code into one code base. * vm.c, vm_eval.c: catch up these changes. * compile.c (new_callinfo): callinfo requires kw_arg parameter. * compile.c (compile_array_): check the last argument Hash object or not. If Hash object and all keys are Symbol literals, they are compiled to keyword arguments. * insns.def (checkkeyword): add new instruction. This instruction check the availability of corresponding keyword. For example, a method "def foo k1: 'v1'; end" is cimpiled to the following instructions. 0000 checkkeyword 2, 0 # check k1 is given. 0003 branchif 9 # if given, jump to address #9 0005 putstring "v1" 0007 setlocal_OP__WC__0 3 # k1 = 'v1' 0009 trace 8 0011 putnil 0012 trace 16 0014 leave * insns.def (opt_send_simple): removed and add new instruction "opt_send_without_block". * parse.y (new_args_tail_gen): reorder variables. Before this patch, a method "def foo(k1: 1, kr1:, k2: 2, **krest, &b)" has parameter variables "k1, kr1, k2, &b, internal_id, krest", but this patch reorders to "kr1, k1, k2, internal_id, krest, &b". (locate a block variable at last) * parse.y (vtable_pop): added. This function remove latest `n' variables from vtable. * iseq.c: catch up iseq data changes. * proc.c: ditto. * class.c (keyword_error): export as rb_keyword_error(). * common.mk: depend vm_args.c for vm.o. * hash.c (rb_hash_has_key): export. * internal.h: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@48239 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2014-11-02 21:02:55 +03:00
if (body->param.lead_num == 1 && last_comma == 0) {
/* {|a|} */
body->param.flags.ambiguous_param0 = TRUE;
}
}
}
}
return COMPILE_OK;
}
static int
iseq_set_local_table(rb_iseq_t *iseq, const ID *tbl)
{
unsigned int size;
if (tbl) {
size = (unsigned int)*tbl;
tbl++;
}
else {
size = 0;
}
if (size > 0) {
ID *ids = (ID *)ALLOC_N(ID, size);
MEMCPY(ids, tbl, ID, size);
iseq->body->local_table = ids;
}
iseq->body->local_table_size = size;
debugs("iseq_set_local_table: %u\n", iseq->body->local_table_size);
return COMPILE_OK;
}
static int
cdhash_cmp(VALUE val, VALUE lit)
{
int tval, tlit;
if (val == lit) {
return 0;
}
else if ((tlit = OBJ_BUILTIN_TYPE(lit)) == -1) {
return val != lit;
}
else if ((tval = OBJ_BUILTIN_TYPE(val)) == -1) {
return -1;
}
else if (tlit != tval) {
return -1;
}
else if (tlit == T_SYMBOL) {
return val != lit;
}
else if (tlit == T_STRING) {
return rb_str_hash_cmp(lit, val);
}
else if (tlit == T_BIGNUM) {
long x = FIX2LONG(rb_big_cmp(lit, val));
/* Given lit and val are both Bignum, x must be -1, 0, 1.
* There is no need to call rb_fix2int here. */
RUBY_ASSERT((x == 1) || (x == 0) || (x == -1));
return (int)x;
}
else if (tlit == T_FLOAT) {
return rb_float_cmp(lit, val);
}
else {
UNREACHABLE_RETURN(-1);
}
}
static st_index_t
cdhash_hash(VALUE a)
{
switch (OBJ_BUILTIN_TYPE(a)) {
case -1:
case T_SYMBOL:
return (st_index_t)a;
case T_STRING:
return rb_str_hash(a);
case T_BIGNUM:
return FIX2LONG(rb_big_hash(a));
case T_FLOAT:
return rb_dbl_long_hash(RFLOAT_VALUE(a));
default:
UNREACHABLE_RETURN(0);
}
}
static const struct st_hash_type cdhash_type = {
cdhash_cmp,
cdhash_hash,
};
struct cdhash_set_label_struct {
VALUE hash;
int pos;
int len;
};
static int
cdhash_set_label_i(VALUE key, VALUE val, VALUE ptr)
{
struct cdhash_set_label_struct *data = (struct cdhash_set_label_struct *)ptr;
LABEL *lobj = (LABEL *)(val & ~1);
rb_hash_aset(data->hash, key, INT2FIX(lobj->position - (data->pos+data->len)));
return ST_CONTINUE;
}
static inline VALUE
get_ivar_ic_value(rb_iseq_t *iseq,ID id)
{
VALUE val;
struct rb_id_table *tbl = ISEQ_COMPILE_DATA(iseq)->ivar_cache_table;
if (tbl) {
if (rb_id_table_lookup(tbl,id,&val)) {
return val;
}
}
else {
tbl = rb_id_table_create(1);
ISEQ_COMPILE_DATA(iseq)->ivar_cache_table = tbl;
}
val = INT2FIX(iseq->body->is_size++);
rb_id_table_insert(tbl,id,val);
return val;
}
#define BADINSN_DUMP(anchor, list, dest) \
dump_disasm_list_with_cursor(FIRST_ELEMENT(anchor), list, dest)
#define BADINSN_ERROR \
(xfree(generated_iseq), \
xfree(insns_info), \
BADINSN_DUMP(anchor, list, NULL), \
COMPILE_ERROR)
static int
fix_sp_depth(rb_iseq_t *iseq, LINK_ANCHOR *const anchor)
{
int stack_max = 0, sp = 0, line = 0;
LINK_ELEMENT *list;
for (list = FIRST_ELEMENT(anchor); list; list = list->next) {
if (list->type == ISEQ_ELEMENT_LABEL) {
LABEL *lobj = (LABEL *)list;
lobj->set = TRUE;
}
}
for (list = FIRST_ELEMENT(anchor); list; list = list->next) {
switch (list->type) {
case ISEQ_ELEMENT_INSN:
{
int j, len, insn;
const char *types;
VALUE *operands;
INSN *iobj = (INSN *)list;
/* update sp */
sp = calc_sp_depth(sp, iobj);
if (sp < 0) {
BADINSN_DUMP(anchor, list, NULL);
COMPILE_ERROR(iseq, iobj->insn_info.line_no,
"argument stack underflow (%d)", sp);
return -1;
}
if (sp > stack_max) {
stack_max = sp;
}
line = iobj->insn_info.line_no;
/* fprintf(stderr, "insn: %-16s, sp: %d\n", insn_name(iobj->insn_id), sp); */
operands = iobj->operands;
insn = iobj->insn_id;
types = insn_op_types(insn);
len = insn_len(insn);
/* operand check */
if (iobj->operand_size != len - 1) {
/* printf("operand size miss! (%d, %d)\n", iobj->operand_size, len); */
BADINSN_DUMP(anchor, list, NULL);
COMPILE_ERROR(iseq, iobj->insn_info.line_no,
"operand size miss! (%d for %d)",
iobj->operand_size, len - 1);
return -1;
}
for (j = 0; types[j]; j++) {
if (types[j] == TS_OFFSET) {
/* label(destination position) */
LABEL *lobj = (LABEL *)operands[j];
if (!lobj->set) {
BADINSN_DUMP(anchor, list, NULL);
COMPILE_ERROR(iseq, iobj->insn_info.line_no,
"unknown label: "LABEL_FORMAT, lobj->label_no);
return -1;
}
if (lobj->sp == -1) {
lobj->sp = sp;
}
}
}
break;
}
case ISEQ_ELEMENT_LABEL:
{
LABEL *lobj = (LABEL *)list;
if (lobj->sp == -1) {
lobj->sp = sp;
}
else {
sp = lobj->sp;
}
break;
}
case ISEQ_ELEMENT_TRACE:
{
/* ignore */
break;
}
case ISEQ_ELEMENT_ADJUST:
{
ADJUST *adjust = (ADJUST *)list;
int orig_sp = sp;
sp = adjust->label ? adjust->label->sp : 0;
if (adjust->line_no != -1 && orig_sp - sp < 0) {
BADINSN_DUMP(anchor, list, NULL);
COMPILE_ERROR(iseq, adjust->line_no,
"iseq_set_sequence: adjust bug %d < %d",
orig_sp, sp);
return -1;
}
break;
}
default:
BADINSN_DUMP(anchor, list, NULL);
COMPILE_ERROR(iseq, line, "unknown list type: %d", list->type);
return -1;
}
}
return stack_max;
}
static int
add_insn_info(struct iseq_insn_info_entry *insns_info, unsigned int *positions,
int insns_info_index, int code_index, const INSN *iobj)
{
if (insns_info_index == 0 ||
insns_info[insns_info_index-1].line_no != iobj->insn_info.line_no ||
insns_info[insns_info_index-1].events != iobj->insn_info.events) {
insns_info[insns_info_index].line_no = iobj->insn_info.line_no;
insns_info[insns_info_index].events = iobj->insn_info.events;
positions[insns_info_index] = code_index;
return TRUE;
}
return FALSE;
}
static int
add_adjust_info(struct iseq_insn_info_entry *insns_info, unsigned int *positions,
int insns_info_index, int code_index, const ADJUST *adjust)
{
if (insns_info_index > 0 ||
insns_info[insns_info_index-1].line_no != adjust->line_no) {
insns_info[insns_info_index].line_no = adjust->line_no;
insns_info[insns_info_index].events = 0;
positions[insns_info_index] = code_index;
return TRUE;
}
return FALSE;
}
/**
ruby insn object list -> raw instruction sequence
*/
static int
iseq_set_sequence(rb_iseq_t *iseq, LINK_ANCHOR *const anchor)
{
struct iseq_insn_info_entry *insns_info;
struct rb_iseq_constant_body *const body = iseq->body;
unsigned int *positions;
LINK_ELEMENT *list;
VALUE *generated_iseq;
rb_event_flag_t events = 0;
long data = 0;
int insn_num, code_index, insns_info_index, sp = 0;
int stack_max = fix_sp_depth(iseq, anchor);
if (stack_max < 0) return COMPILE_NG;
/* fix label position */
list = FIRST_ELEMENT(anchor);
insn_num = code_index = 0;
while (list) {
switch (list->type) {
case ISEQ_ELEMENT_INSN:
{
INSN *iobj = (INSN *)list;
/* update sp */
sp = calc_sp_depth(sp, iobj);
insn_num++;
events = iobj->insn_info.events |= events;
if (ISEQ_COVERAGE(iseq)) {
if (ISEQ_LINE_COVERAGE(iseq) && (events & RUBY_EVENT_COVERAGE_LINE) &&
!(rb_get_coverage_mode() & COVERAGE_TARGET_ONESHOT_LINES)) {
int line = iobj->insn_info.line_no;
if (line >= 1) {
RARRAY_ASET(ISEQ_LINE_COVERAGE(iseq), line - 1, INT2FIX(0));
}
}
if (ISEQ_BRANCH_COVERAGE(iseq) && (events & RUBY_EVENT_COVERAGE_BRANCH)) {
while (RARRAY_LEN(ISEQ_PC2BRANCHINDEX(iseq)) <= code_index) {
rb_ary_push(ISEQ_PC2BRANCHINDEX(iseq), Qnil);
}
RARRAY_ASET(ISEQ_PC2BRANCHINDEX(iseq), code_index, INT2FIX(data));
}
}
code_index += insn_data_length(iobj);
events = 0;
data = 0;
break;
}
case ISEQ_ELEMENT_LABEL:
{
LABEL *lobj = (LABEL *)list;
lobj->position = code_index;
sp = lobj->sp;
break;
}
case ISEQ_ELEMENT_TRACE:
{
TRACE *trace = (TRACE *)list;
events |= trace->event;
if (trace->event & RUBY_EVENT_COVERAGE_BRANCH) data = trace->data;
break;
}
case ISEQ_ELEMENT_ADJUST:
{
ADJUST *adjust = (ADJUST *)list;
if (adjust->line_no != -1) {
int orig_sp = sp;
sp = adjust->label ? adjust->label->sp : 0;
if (orig_sp - sp > 0) {
if (orig_sp - sp > 1) code_index++; /* 1 operand */
code_index++; /* insn */
insn_num++;
}
}
break;
}
default: break;
}
list = list->next;
}
/* make instruction sequence */
generated_iseq = ALLOC_N(VALUE, code_index);
insns_info = ALLOC_N(struct iseq_insn_info_entry, insn_num);
positions = ALLOC_N(unsigned int, insn_num);
body->is_entries = ZALLOC_N(union iseq_inline_storage_entry, body->is_size);
body->call_data =
rb_xcalloc_mul_add_mul(
sizeof(struct rb_call_data), body->ci_size,
sizeof(struct rb_kwarg_call_data), body->ci_kw_size);
ISEQ_COMPILE_DATA(iseq)->ci_index = ISEQ_COMPILE_DATA(iseq)->ci_kw_index = 0;
list = FIRST_ELEMENT(anchor);
insns_info_index = code_index = sp = 0;
while (list) {
switch (list->type) {
case ISEQ_ELEMENT_INSN:
{
int j, len, insn;
const char *types;
VALUE *operands;
INSN *iobj = (INSN *)list;
/* update sp */
sp = calc_sp_depth(sp, iobj);
/* fprintf(stderr, "insn: %-16s, sp: %d\n", insn_name(iobj->insn_id), sp); */
operands = iobj->operands;
insn = iobj->insn_id;
generated_iseq[code_index] = insn;
types = insn_op_types(insn);
len = insn_len(insn);
for (j = 0; types[j]; j++) {
char type = types[j];
/* printf("--> [%c - (%d-%d)]\n", type, k, j); */
switch (type) {
case TS_OFFSET:
{
/* label(destination position) */
LABEL *lobj = (LABEL *)operands[j];
generated_iseq[code_index + 1 + j] = lobj->position - (code_index + len);
break;
}
case TS_CDHASH:
{
VALUE map = operands[j];
struct cdhash_set_label_struct data;
data.hash = map;
data.pos = code_index;
data.len = len;
rb_hash_foreach(map, cdhash_set_label_i, (VALUE)&data);
rb_hash_rehash(map);
2015-07-22 01:52:59 +03:00
freeze_hide_obj(map);
generated_iseq[code_index + 1 + j] = map;
RB_OBJ_WRITTEN(iseq, Qundef, map);
FL_SET(iseq, ISEQ_MARKABLE_ISEQ);
break;
}
case TS_LINDEX:
case TS_NUM: /* ulong */
generated_iseq[code_index + 1 + j] = FIX2INT(operands[j]);
break;
case TS_VALUE: /* VALUE */
case TS_ISEQ: /* iseq */
{
VALUE v = operands[j];
generated_iseq[code_index + 1 + j] = v;
/* to mark ruby object */
if (!SPECIAL_CONST_P(v)) {
RB_OBJ_WRITTEN(iseq, Qundef, v);
FL_SET(iseq, ISEQ_MARKABLE_ISEQ);
}
break;
}
case TS_ISE: /* inline storage entry */
/* Treated as an IC, but may contain a markable VALUE */
FL_SET(iseq, ISEQ_MARKABLE_ISEQ);
2019-07-14 18:31:07 +03:00
/* fall through */
case TS_IC: /* inline cache */
{
unsigned int ic_index = FIX2UINT(operands[j]);
IC ic = (IC)&body->is_entries[ic_index];
if (UNLIKELY(ic_index >= body->is_size)) {
BADINSN_DUMP(anchor, &iobj->link, 0);
COMPILE_ERROR(iseq, iobj->insn_info.line_no,
"iseq_set_sequence: ic_index overflow: index: %d, size: %d",
ic_index, body->is_size);
}
generated_iseq[code_index + 1 + j] = (VALUE)ic;
break;
}
case TS_CALLDATA:
{
struct rb_call_info *source_ci = (struct rb_call_info *)operands[j];
struct rb_call_data *cd;
if (source_ci->flag & VM_CALL_KWARG) {
struct rb_kwarg_call_data *kw_calls = (struct rb_kwarg_call_data *)&body->call_data[body->ci_size];
struct rb_kwarg_call_data *cd_kw = &kw_calls[ISEQ_COMPILE_DATA(iseq)->ci_kw_index++];
cd_kw->ci_kw = *((struct rb_call_info_with_kwarg *)source_ci);
cd = (struct rb_call_data *)cd_kw;
assert(ISEQ_COMPILE_DATA(iseq)->ci_kw_index <= body->ci_kw_size);
}
else {
cd = &body->call_data[ISEQ_COMPILE_DATA(iseq)->ci_index++];
cd->ci = *source_ci;
assert(ISEQ_COMPILE_DATA(iseq)->ci_index <= body->ci_size);
}
generated_iseq[code_index + 1 + j] = (VALUE)cd;
break;
}
case TS_ID: /* ID */
generated_iseq[code_index + 1 + j] = SYM2ID(operands[j]);
break;
case TS_GENTRY:
{
struct rb_global_entry *entry =
(struct rb_global_entry *)(operands[j] & (~1));
generated_iseq[code_index + 1 + j] = (VALUE)entry;
}
break;
case TS_FUNCPTR:
generated_iseq[code_index + 1 + j] = operands[j];
break;
default:
BADINSN_ERROR(iseq, iobj->insn_info.line_no,
"unknown operand type: %c", type);
return COMPILE_NG;
}
}
if (add_insn_info(insns_info, positions, insns_info_index, code_index, iobj)) insns_info_index++;
code_index += len;
break;
}
case ISEQ_ELEMENT_LABEL:
{
LABEL *lobj = (LABEL *)list;
sp = lobj->sp;
break;
}
case ISEQ_ELEMENT_ADJUST:
{
ADJUST *adjust = (ADJUST *)list;
int orig_sp = sp;
if (adjust->label) {
sp = adjust->label->sp;
}
else {
sp = 0;
}
if (adjust->line_no != -1) {
const int diff = orig_sp - sp;
if (diff > 0) {
if (add_adjust_info(insns_info, positions, insns_info_index, code_index, adjust)) insns_info_index++;
}
if (diff > 1) {
generated_iseq[code_index++] = BIN(adjuststack);
generated_iseq[code_index++] = orig_sp - sp;
}
else if (diff == 1) {
generated_iseq[code_index++] = BIN(pop);
}
else if (diff < 0) {
int label_no = adjust->label ? adjust->label->label_no : -1;
xfree(generated_iseq);
xfree(insns_info);
xfree(positions);
debug_list(anchor);
COMPILE_ERROR(iseq, adjust->line_no,
"iseq_set_sequence: adjust bug to %d %d < %d",
label_no, orig_sp, sp);
return COMPILE_NG;
}
}
break;
}
default:
/* ignore */
break;
}
list = list->next;
}
body->iseq_encoded = (void *)generated_iseq;
body->iseq_size = code_index;
body->stack_max = stack_max;
/* get rid of memory leak when REALLOC failed */
body->insns_info.body = insns_info;
body->insns_info.positions = positions;
REALLOC_N(insns_info, struct iseq_insn_info_entry, insns_info_index);
body->insns_info.body = insns_info;
REALLOC_N(positions, unsigned int, insns_info_index);
body->insns_info.positions = positions;
body->insns_info.size = insns_info_index;
return COMPILE_OK;
}
static int
label_get_position(LABEL *lobj)
{
return lobj->position;
}
static int
label_get_sp(LABEL *lobj)
{
return lobj->sp;
}
static int
iseq_set_exception_table(rb_iseq_t *iseq)
{
const VALUE *tptr, *ptr;
unsigned int tlen, i;
struct iseq_catch_table_entry *entry;
if (NIL_P(ISEQ_COMPILE_DATA(iseq)->catch_table_ary)) goto no_catch_table;
tlen = (int)RARRAY_LEN(ISEQ_COMPILE_DATA(iseq)->catch_table_ary);
tptr = RARRAY_CONST_PTR_TRANSIENT(ISEQ_COMPILE_DATA(iseq)->catch_table_ary);
if (tlen > 0) {
struct iseq_catch_table *table = xmalloc(iseq_catch_table_bytes(tlen));
table->size = tlen;
for (i = 0; i < table->size; i++) {
ptr = RARRAY_CONST_PTR_TRANSIENT(tptr[i]);
entry = UNALIGNED_MEMBER_PTR(table, entries[i]);
entry->type = (enum catch_type)(ptr[0] & 0xffff);
entry->start = label_get_position((LABEL *)(ptr[1] & ~1));
entry->end = label_get_position((LABEL *)(ptr[2] & ~1));
entry->iseq = (rb_iseq_t *)ptr[3];
RB_OBJ_WRITTEN(iseq, Qundef, entry->iseq);
/* stack depth */
if (ptr[4]) {
LABEL *lobj = (LABEL *)(ptr[4] & ~1);
entry->cont = label_get_position(lobj);
entry->sp = label_get_sp(lobj);
/* TODO: Dirty Hack! Fix me */
if (entry->type == CATCH_TYPE_RESCUE ||
entry->type == CATCH_TYPE_BREAK ||
entry->type == CATCH_TYPE_NEXT) {
entry->sp--;
}
}
else {
entry->cont = 0;
}
}
iseq->body->catch_table = table;
RB_OBJ_WRITE(iseq, &ISEQ_COMPILE_DATA(iseq)->catch_table_ary, 0); /* free */
}
else {
no_catch_table:
iseq->body->catch_table = NULL;
}
return COMPILE_OK;
}
/*
* set optional argument table
* def foo(a, b=expr1, c=expr2)
* =>
* b:
* expr1
* c:
* expr2
*/
static int
iseq_set_optargs_table(rb_iseq_t *iseq)
{
int i;
VALUE *opt_table = (VALUE *)iseq->body->param.opt_table;
2015-07-22 01:52:59 +03:00
if (iseq->body->param.flags.has_opt) {
for (i = 0; i < iseq->body->param.opt_num + 1; i++) {
opt_table[i] = label_get_position((LABEL *)opt_table[i]);
}
}
return COMPILE_OK;
}
static LINK_ELEMENT *
get_destination_insn(INSN *iobj)
{
LABEL *lobj = (LABEL *)OPERAND_AT(iobj, 0);
LINK_ELEMENT *list;
rb_event_flag_t events = 0;
list = lobj->link.next;
while (list) {
switch (list->type) {
case ISEQ_ELEMENT_INSN:
case ISEQ_ELEMENT_ADJUST:
goto found;
case ISEQ_ELEMENT_LABEL:
/* ignore */
break;
case ISEQ_ELEMENT_TRACE:
{
TRACE *trace = (TRACE *)list;
events |= trace->event;
}
break;
default: break;
}
list = list->next;
}
found:
if (list && IS_INSN(list)) {
INSN *iobj = (INSN *)list;
iobj->insn_info.events |= events;
}
return list;
}
static LINK_ELEMENT *
get_next_insn(INSN *iobj)
{
LINK_ELEMENT *list = iobj->link.next;
while (list) {
if (IS_INSN(list) || IS_ADJUST(list)) {
return list;
}
list = list->next;
}
return 0;
}
static LINK_ELEMENT *
get_prev_insn(INSN *iobj)
{
LINK_ELEMENT *list = iobj->link.prev;
while (list) {
if (IS_INSN(list) || IS_ADJUST(list)) {
return list;
}
list = list->prev;
}
return 0;
}
static void
unref_destination(INSN *iobj, int pos)
{
LABEL *lobj = (LABEL *)OPERAND_AT(iobj, pos);
--lobj->refcnt;
if (!lobj->refcnt) ELEM_REMOVE(&lobj->link);
}
static void
replace_destination(INSN *dobj, INSN *nobj)
{
VALUE n = OPERAND_AT(nobj, 0);
LABEL *dl = (LABEL *)OPERAND_AT(dobj, 0);
LABEL *nl = (LABEL *)n;
--dl->refcnt;
++nl->refcnt;
OPERAND_AT(dobj, 0) = n;
if (!dl->refcnt) ELEM_REMOVE(&dl->link);
}
static LABEL*
find_destination(INSN *i)
{
int pos, len = insn_len(i->insn_id);
for (pos = 0; pos < len; ++pos) {
if (insn_op_types(i->insn_id)[pos] == TS_OFFSET) {
return (LABEL *)OPERAND_AT(i, pos);
}
}
return 0;
}
static int
remove_unreachable_chunk(rb_iseq_t *iseq, LINK_ELEMENT *i)
{
LINK_ELEMENT *first = i, *end;
int *unref_counts = 0, nlabels = ISEQ_COMPILE_DATA(iseq)->label_no;
if (!i) return 0;
unref_counts = ALLOCA_N(int, nlabels);
MEMZERO(unref_counts, int, nlabels);
end = i;
do {
LABEL *lab;
if (IS_INSN(i)) {
if (IS_INSN_ID(i, leave)) {
end = i;
break;
}
else if ((lab = find_destination((INSN *)i)) != 0) {
if (lab->unremovable) break;
unref_counts[lab->label_no]++;
}
}
else if (IS_LABEL(i)) {
lab = (LABEL *)i;
if (lab->unremovable) return 0;
if (lab->refcnt > unref_counts[lab->label_no]) {
if (i == first) return 0;
break;
}
continue;
}
else if (IS_TRACE(i)) {
/* do nothing */
}
else if (IS_ADJUST(i)) {
LABEL *dest = ((ADJUST *)i)->label;
if (dest && dest->unremovable) return 0;
}
end = i;
} while ((i = i->next) != 0);
i = first;
do {
if (IS_INSN(i)) {
struct rb_iseq_constant_body *body = iseq->body;
VALUE insn = INSN_OF(i);
int pos, len = insn_len(insn);
for (pos = 0; pos < len; ++pos) {
switch (insn_op_types(insn)[pos]) {
case TS_OFFSET:
unref_destination((INSN *)i, pos);
break;
case TS_CALLDATA:
if (((struct rb_call_info *)OPERAND_AT(i, pos))->flag & VM_CALL_KWARG)
--(body->ci_kw_size);
else
--(body->ci_size);
break;
}
}
}
ELEM_REMOVE(i);
} while ((i != end) && (i = i->next) != 0);
return 1;
}
static int
iseq_pop_newarray(rb_iseq_t *iseq, INSN *iobj)
{
switch (OPERAND_AT(iobj, 0)) {
case INT2FIX(0): /* empty array */
ELEM_REMOVE(&iobj->link);
return TRUE;
case INT2FIX(1): /* single element array */
ELEM_REMOVE(&iobj->link);
return FALSE;
default:
iobj->insn_id = BIN(adjuststack);
return TRUE;
}
}
static int
same_debug_pos_p(LINK_ELEMENT *iobj1, LINK_ELEMENT *iobj2)
{
VALUE debug1 = OPERAND_AT(iobj1, 0);
VALUE debug2 = OPERAND_AT(iobj2, 0);
if (debug1 == debug2) return TRUE;
if (!RB_TYPE_P(debug1, T_ARRAY)) return FALSE;
if (!RB_TYPE_P(debug2, T_ARRAY)) return FALSE;
if (RARRAY_LEN(debug1) != 2) return FALSE;
if (RARRAY_LEN(debug2) != 2) return FALSE;
if (RARRAY_AREF(debug1, 0) != RARRAY_AREF(debug2, 0)) return FALSE;
if (RARRAY_AREF(debug1, 1) != RARRAY_AREF(debug2, 1)) return FALSE;
return TRUE;
}
static int
is_frozen_putstring(INSN *insn, VALUE *op)
{
if (IS_INSN_ID(insn, putstring)) {
*op = OPERAND_AT(insn, 0);
return 1;
}
else if (IS_INSN_ID(insn, putobject)) { /* frozen_string_literal */
*op = OPERAND_AT(insn, 0);
return RB_TYPE_P(*op, T_STRING);
}
return 0;
}
static int
optimize_checktype(rb_iseq_t *iseq, INSN *iobj)
{
/*
* putobject obj
* dup
* checktype T_XXX
* branchif l1
* l2:
* ...
* l1:
*
* => obj is a T_XXX
*
* putobject obj (T_XXX)
* jump L1
* L1:
*
* => obj is not a T_XXX
*
* putobject obj (T_XXX)
* jump L2
* L2:
*/
int line;
INSN *niobj, *ciobj, *dup = 0;
LABEL *dest = 0;
VALUE type;
switch (INSN_OF(iobj)) {
case BIN(putstring):
type = INT2FIX(T_STRING);
break;
case BIN(putnil):
type = INT2FIX(T_NIL);
break;
case BIN(putobject):
type = INT2FIX(TYPE(OPERAND_AT(iobj, 0)));
break;
default: return FALSE;
}
ciobj = (INSN *)get_next_insn(iobj);
if (IS_INSN_ID(ciobj, jump)) {
ciobj = (INSN *)get_next_insn((INSN*)OPERAND_AT(ciobj, 0));
}
if (IS_INSN_ID(ciobj, dup)) {
ciobj = (INSN *)get_next_insn(dup = ciobj);
}
if (!ciobj || !IS_INSN_ID(ciobj, checktype)) return FALSE;
niobj = (INSN *)get_next_insn(ciobj);
if (!niobj) {
no_branch:
/* TODO: putobject true/false */
return FALSE;
}
switch (INSN_OF(niobj)) {
case BIN(branchif):
if (OPERAND_AT(ciobj, 0) == type) {
dest = (LABEL *)OPERAND_AT(niobj, 0);
}
break;
case BIN(branchunless):
if (OPERAND_AT(ciobj, 0) != type) {
dest = (LABEL *)OPERAND_AT(niobj, 0);
}
break;
default:
goto no_branch;
}
line = ciobj->insn_info.line_no;
if (!dest) {
if (niobj->link.next && IS_LABEL(niobj->link.next)) {
dest = (LABEL *)niobj->link.next; /* reuse label */
}
else {
dest = NEW_LABEL(line);
ELEM_INSERT_NEXT(&niobj->link, &dest->link);
}
}
INSERT_AFTER_INSN1(iobj, line, jump, dest);
LABEL_REF(dest);
if (!dup) INSERT_AFTER_INSN(iobj, line, pop);
return TRUE;
}
static int
iseq_peephole_optimize(rb_iseq_t *iseq, LINK_ELEMENT *list, const int do_tailcallopt)
{
INSN *const iobj = (INSN *)list;
again:
optimize_checktype(iseq, iobj);
if (IS_INSN_ID(iobj, jump)) {
INSN *niobj, *diobj, *piobj;
diobj = (INSN *)get_destination_insn(iobj);
niobj = (INSN *)get_next_insn(iobj);
if (diobj == niobj) {
/*
* jump LABEL
* LABEL:
* =>
* LABEL:
*/
unref_destination(iobj, 0);
ELEM_REMOVE(&iobj->link);
return COMPILE_OK;
}
else if (iobj != diobj && IS_INSN_ID(diobj, jump) &&
OPERAND_AT(iobj, 0) != OPERAND_AT(diobj, 0)) {
/*
* useless jump elimination:
* jump LABEL1
* ...
* LABEL1:
* jump LABEL2
*
* => in this case, first jump instruction should jump to
* LABEL2 directly
*/
replace_destination(iobj, diobj);
remove_unreachable_chunk(iseq, iobj->link.next);
goto again;
}
else if (IS_INSN_ID(diobj, leave)) {
INSN *pop;
/*
* jump LABEL
* ...
* LABEL:
* leave
* =>
* leave
* pop
* ...
* LABEL:
* leave
*/
/* replace */
unref_destination(iobj, 0);
iobj->insn_id = BIN(leave);
iobj->operand_size = 0;
iobj->insn_info = diobj->insn_info;
/* adjust stack depth */
pop = new_insn_body(iseq, diobj->insn_info.line_no, BIN(pop), 0);
ELEM_INSERT_NEXT(&iobj->link, &pop->link);
goto again;
}
else if (IS_INSN(iobj->link.prev) &&
(piobj = (INSN *)iobj->link.prev) &&
(IS_INSN_ID(piobj, branchif) ||
IS_INSN_ID(piobj, branchunless))) {
INSN *pdiobj = (INSN *)get_destination_insn(piobj);
if (niobj == pdiobj) {
int refcnt = IS_LABEL(piobj->link.next) ?
((LABEL *)piobj->link.next)->refcnt : 0;
/*
* useless jump elimination (if/unless destination):
* if L1
* jump L2
* L1:
* ...
* L2:
*
* ==>
* unless L2
* L1:
* ...
* L2:
*/
piobj->insn_id = (IS_INSN_ID(piobj, branchif))
? BIN(branchunless) : BIN(branchif);
replace_destination(piobj, iobj);
if (refcnt <= 1) {
ELEM_REMOVE(&iobj->link);
}
else {
/* TODO: replace other branch destinations too */
}
return COMPILE_OK;
}
else if (diobj == pdiobj) {
/*
* useless jump elimination (if/unless before jump):
* L1:
* ...
* if L1
* jump L1
*
* ==>
* L1:
* ...
* pop
* jump L1
*/
INSN *popiobj = new_insn_core(iseq, iobj->insn_info.line_no,
BIN(pop), 0, 0);
ELEM_REPLACE(&piobj->link, &popiobj->link);
}
}
if (remove_unreachable_chunk(iseq, iobj->link.next)) {
goto again;
}
}
/*
* putstring "beg"
* putstring "end"
* newrange excl
*
* ==>
*
* putobject "beg".."end"
*/
if (IS_INSN_ID(iobj, checkmatch)) {
INSN *range = (INSN *)get_prev_insn(iobj);
INSN *beg, *end;
VALUE str_beg, str_end;
if (range && IS_INSN_ID(range, newrange) &&
(end = (INSN *)get_prev_insn(range)) != 0 &&
is_frozen_putstring(end, &str_end) &&
(beg = (INSN *)get_prev_insn(end)) != 0 &&
is_frozen_putstring(beg, &str_beg)) {
int excl = FIX2INT(OPERAND_AT(range, 0));
VALUE lit_range = rb_range_new(str_beg, str_end, excl);
ELEM_REMOVE(&beg->link);
ELEM_REMOVE(&end->link);
range->insn_id = BIN(putobject);
OPERAND_AT(range, 0) = lit_range;
RB_OBJ_WRITTEN(iseq, Qundef, lit_range);
}
}
if (IS_INSN_ID(iobj, leave)) {
remove_unreachable_chunk(iseq, iobj->link.next);
}
if (IS_INSN_ID(iobj, branchif) ||
IS_INSN_ID(iobj, branchnil) ||
IS_INSN_ID(iobj, branchunless)) {
/*
* if L1
* ...
* L1:
* jump L2
* =>
* if L2
*/
INSN *nobj = (INSN *)get_destination_insn(iobj);
INSN *pobj = (INSN *)iobj->link.prev;
int prev_dup = 0;
if (pobj) {
if (!IS_INSN(&pobj->link))
pobj = 0;
else if (IS_INSN_ID(pobj, dup))
prev_dup = 1;
}
for (;;) {
if (IS_INSN_ID(nobj, jump)) {
replace_destination(iobj, nobj);
}
else if (prev_dup && IS_INSN_ID(nobj, dup) &&
!!(nobj = (INSN *)nobj->link.next) &&
/* basic blocks, with no labels in the middle */
nobj->insn_id == iobj->insn_id) {
/*
* dup
* if L1
* ...
* L1:
* dup
* if L2
* =>
* dup
* if L2
* ...
* L1:
* dup
* if L2
*/
replace_destination(iobj, nobj);
}
else if (pobj) {
/*
* putnil
* if L1
* =>
* # nothing
*
* putobject true
* if L1
* =>
* jump L1
*
* putstring ".."
* if L1
* =>
* jump L1
*
* putstring ".."
* dup
* if L1
* =>
* putstring ".."
* jump L1
*
*/
int cond;
if (prev_dup && IS_INSN(pobj->link.prev)) {
pobj = (INSN *)pobj->link.prev;
}
if (IS_INSN_ID(pobj, putobject)) {
cond = (IS_INSN_ID(iobj, branchif) ?
OPERAND_AT(pobj, 0) != Qfalse :
IS_INSN_ID(iobj, branchunless) ?
OPERAND_AT(pobj, 0) == Qfalse :
FALSE);
}
else if (IS_INSN_ID(pobj, putstring) ||
IS_INSN_ID(pobj, duparray) ||
IS_INSN_ID(pobj, newarray)) {
cond = IS_INSN_ID(iobj, branchif);
}
else if (IS_INSN_ID(pobj, putnil)) {
cond = !IS_INSN_ID(iobj, branchif);
}
else break;
if (prev_dup || !IS_INSN_ID(pobj, newarray)) {
ELEM_REMOVE(iobj->link.prev);
}
else if (!iseq_pop_newarray(iseq, pobj)) {
pobj = new_insn_core(iseq, pobj->insn_info.line_no, BIN(pop), 0, NULL);
ELEM_INSERT_PREV(&iobj->link, &pobj->link);
}
if (cond) {
if (prev_dup) {
pobj = new_insn_core(iseq, pobj->insn_info.line_no, BIN(putnil), 0, NULL);
ELEM_INSERT_NEXT(&iobj->link, &pobj->link);
}
iobj->insn_id = BIN(jump);
goto again;
}
else {
unref_destination(iobj, 0);
ELEM_REMOVE(&iobj->link);
}
break;
}
else break;
nobj = (INSN *)get_destination_insn(nobj);
}
}
if (IS_INSN_ID(iobj, pop)) {
/*
* putself / putnil / putobject obj / putstring "..."
* pop
* =>
* # do nothing
*/
LINK_ELEMENT *prev = iobj->link.prev;
if (IS_INSN(prev)) {
enum ruby_vminsn_type previ = ((INSN *)prev)->insn_id;
if (previ == BIN(putobject) || previ == BIN(putnil) ||
previ == BIN(putself) || previ == BIN(putstring) ||
previ == BIN(dup) ||
previ == BIN(getlocal) ||
previ == BIN(getblockparam) ||
previ == BIN(getblockparamproxy) ||
/* getinstancevariable may issue a warning */
previ == BIN(duparray)) {
/* just push operand or static value and pop soon, no
* side effects */
ELEM_REMOVE(prev);
ELEM_REMOVE(&iobj->link);
}
else if (previ == BIN(newarray) && iseq_pop_newarray(iseq, (INSN*)prev)) {
ELEM_REMOVE(&iobj->link);
}
else if (previ == BIN(concatarray)) {
INSN *piobj = (INSN *)prev;
INSERT_BEFORE_INSN1(piobj, piobj->insn_info.line_no, splatarray, Qfalse);
INSN_OF(piobj) = BIN(pop);
}
else if (previ == BIN(concatstrings)) {
if (OPERAND_AT(prev, 0) == INT2FIX(1)) {
ELEM_REMOVE(prev);
}
else {
ELEM_REMOVE(&iobj->link);
INSN_OF(prev) = BIN(adjuststack);
}
}
}
}
if (IS_INSN_ID(iobj, newarray) ||
IS_INSN_ID(iobj, duparray) ||
IS_INSN_ID(iobj, expandarray) ||
IS_INSN_ID(iobj, concatarray) ||
IS_INSN_ID(iobj, splatarray) ||
0) {
/*
* newarray N
* splatarray
* =>
* newarray N
* newarray always puts an array
*/
LINK_ELEMENT *next = iobj->link.next;
if (IS_INSN(next) && IS_INSN_ID(next, splatarray)) {
/* remove splatarray following always-array insn */
ELEM_REMOVE(next);
}
}
if (IS_INSN_ID(iobj, tostring)) {
LINK_ELEMENT *next = iobj->link.next;
/*
* tostring
* concatstrings 1
* =>
* tostring
*/
if (IS_INSN(next) && IS_INSN_ID(next, concatstrings) &&
OPERAND_AT(next, 0) == INT2FIX(1)) {
ELEM_REMOVE(next);
}
}
if (IS_INSN_ID(iobj, putstring) ||
(IS_INSN_ID(iobj, putobject) && RB_TYPE_P(OPERAND_AT(iobj, 0), T_STRING))) {
/*
* putstring ""
* concatstrings N
* =>
* concatstrings N-1
*/
if (IS_NEXT_INSN_ID(&iobj->link, concatstrings) &&
RSTRING_LEN(OPERAND_AT(iobj, 0)) == 0) {
INSN *next = (INSN *)iobj->link.next;
if ((OPERAND_AT(next, 0) = FIXNUM_INC(OPERAND_AT(next, 0), -1)) == INT2FIX(1)) {
ELEM_REMOVE(&next->link);
}
ELEM_REMOVE(&iobj->link);
}
}
if (IS_INSN_ID(iobj, concatstrings)) {
/*
* concatstrings N
* concatstrings M
* =>
* concatstrings N+M-1
*/
LINK_ELEMENT *next = iobj->link.next, *freeze = 0;
INSN *jump = 0;
if (IS_INSN(next) && IS_INSN_ID(next, freezestring))
next = (freeze = next)->next;
if (IS_INSN(next) && IS_INSN_ID(next, jump))
next = get_destination_insn(jump = (INSN *)next);
if (IS_INSN(next) && IS_INSN_ID(next, concatstrings)) {
int n = FIX2INT(OPERAND_AT(iobj, 0)) + FIX2INT(OPERAND_AT(next, 0)) - 1;
OPERAND_AT(iobj, 0) = INT2FIX(n);
if (jump) {
LABEL *label = ((LABEL *)OPERAND_AT(jump, 0));
if (!--label->refcnt) {
ELEM_REMOVE(&label->link);
}
else {
label = NEW_LABEL(0);
OPERAND_AT(jump, 0) = (VALUE)label;
}
label->refcnt++;
if (freeze && IS_NEXT_INSN_ID(next, freezestring)) {
if (same_debug_pos_p(freeze, next->next)) {
ELEM_REMOVE(freeze);
}
else {
next = next->next;
}
}
ELEM_INSERT_NEXT(next, &label->link);
CHECK(iseq_peephole_optimize(iseq, get_next_insn(jump), do_tailcallopt));
}
else {
if (freeze) ELEM_REMOVE(freeze);
ELEM_REMOVE(next);
}
}
}
if (IS_INSN_ID(iobj, freezestring) &&
NIL_P(OPERAND_AT(iobj, 0)) &&
IS_NEXT_INSN_ID(&iobj->link, send)) {
INSN *niobj = (INSN *)iobj->link.next;
struct rb_call_info *ci = (struct rb_call_info *)OPERAND_AT(niobj, 0);
/*
* freezestring nil # no debug_info
* send <:+@, 0, ARG_SIMPLE> # :-@, too
* =>
* send <:+@, 0, ARG_SIMPLE> # :-@, too
*/
if ((ci->mid == idUPlus || ci->mid == idUMinus) &&
(ci->flag & VM_CALL_ARGS_SIMPLE) &&
ci->orig_argc == 0) {
ELEM_REMOVE(list);
return COMPILE_OK;
}
}
if (do_tailcallopt &&
(IS_INSN_ID(iobj, send) ||
IS_INSN_ID(iobj, opt_aref_with) ||
IS_INSN_ID(iobj, opt_aset_with) ||
IS_INSN_ID(iobj, invokesuper))) {
/*
* send ...
* leave
* =>
* send ..., ... | VM_CALL_TAILCALL, ...
* leave # unreachable
*/
INSN *piobj = NULL;
if (iobj->link.next) {
LINK_ELEMENT *next = iobj->link.next;
do {
if (!IS_INSN(next)) {
next = next->next;
continue;
}
switch (INSN_OF(next)) {
case BIN(nop):
next = next->next;
break;
case BIN(jump):
/* if cond
* return tailcall
* end
*/
next = get_destination_insn((INSN *)next);
break;
case BIN(leave):
piobj = iobj;
2019-07-15 01:59:11 +03:00
/* fall through */
default:
next = NULL;
break;
}
} while (next);
}
if (piobj) {
struct rb_call_info *ci = (struct rb_call_info *)OPERAND_AT(piobj, 0);
if (IS_INSN_ID(piobj, send) || IS_INSN_ID(piobj, invokesuper)) {
if (OPERAND_AT(piobj, 1) == 0) { /* no blockiseq */
ci->flag |= VM_CALL_TAILCALL;
}
}
else {
ci->flag |= VM_CALL_TAILCALL;
}
}
}
if (IS_INSN_ID(iobj, dup)) {
if (IS_NEXT_INSN_ID(&iobj->link, setlocal)) {
LINK_ELEMENT *set1 = iobj->link.next, *set2 = NULL;
if (IS_NEXT_INSN_ID(set1, setlocal)) {
set2 = set1->next;
if (OPERAND_AT(set1, 0) == OPERAND_AT(set2, 0) &&
OPERAND_AT(set1, 1) == OPERAND_AT(set2, 1)) {
ELEM_REMOVE(set1);
ELEM_REMOVE(&iobj->link);
}
}
else if (IS_NEXT_INSN_ID(set1, dup) &&
IS_NEXT_INSN_ID(set1->next, setlocal)) {
set2 = set1->next->next;
if (OPERAND_AT(set1, 0) == OPERAND_AT(set2, 0) &&
OPERAND_AT(set1, 1) == OPERAND_AT(set2, 1)) {
ELEM_REMOVE(set1->next);
ELEM_REMOVE(set2);
}
}
}
}
if (IS_INSN_ID(iobj, getlocal)) {
LINK_ELEMENT *niobj = &iobj->link;
if (IS_NEXT_INSN_ID(niobj, dup)) {
niobj = niobj->next;
}
if (IS_NEXT_INSN_ID(niobj, setlocal)) {
LINK_ELEMENT *set1 = niobj->next;
if (OPERAND_AT(iobj, 0) == OPERAND_AT(set1, 0) &&
OPERAND_AT(iobj, 1) == OPERAND_AT(set1, 1)) {
ELEM_REMOVE(set1);
ELEM_REMOVE(niobj);
}
}
}
return COMPILE_OK;
}
static int
insn_set_specialized_instruction(rb_iseq_t *iseq, INSN *iobj, int insn_id)
{
iobj->insn_id = insn_id;
iobj->operand_size = insn_len(insn_id) - 1;
if (insn_id == BIN(opt_neq)) {
VALUE *old_operands = iobj->operands;
iobj->operand_size = 2;
iobj->operands = compile_data_alloc2(iseq, iobj->operand_size, sizeof(VALUE));
iobj->operands[0] = (VALUE)new_callinfo(iseq, idEq, 1, 0, NULL, FALSE);
iobj->operands[1] = old_operands[0];
}
return COMPILE_OK;
}
static int
iseq_specialized_instruction(rb_iseq_t *iseq, INSN *iobj)
{
if (IS_INSN_ID(iobj, newarray) && iobj->link.next &&
IS_INSN(iobj->link.next)) {
/*
* [a, b, ...].max/min -> a, b, c, opt_newarray_max/min
*/
INSN *niobj = (INSN *)iobj->link.next;
if (IS_INSN_ID(niobj, send)) {
struct rb_call_info *ci = (struct rb_call_info *)OPERAND_AT(niobj, 0);
if ((ci->flag & VM_CALL_ARGS_SIMPLE) && ci->orig_argc == 0) {
switch (ci->mid) {
case idMax:
iobj->insn_id = BIN(opt_newarray_max);
ELEM_REMOVE(&niobj->link);
return COMPILE_OK;
case idMin:
iobj->insn_id = BIN(opt_newarray_min);
ELEM_REMOVE(&niobj->link);
return COMPILE_OK;
}
}
}
}
if (IS_INSN_ID(iobj, send)) {
struct rb_call_info *ci = (struct rb_call_info *)OPERAND_AT(iobj, 0);
const rb_iseq_t *blockiseq = (rb_iseq_t *)OPERAND_AT(iobj, 1);
#define SP_INSN(opt) insn_set_specialized_instruction(iseq, iobj, BIN(opt_##opt))
* rewrite method/block parameter fitting logic to optimize keyword arguments/parameters and a splat argument. [Feature #10440] (Details are described in this ticket) Most of complex part is moved to vm_args.c. Now, ISeq#to_a does not catch up new instruction format. * vm_core.h: change iseq data structures. * introduce rb_call_info_kw_arg_t to represent keyword arguments. * add rb_call_info_t::kw_arg. * rename rb_iseq_t::arg_post_len to rb_iseq_t::arg_post_num. * rename rb_iseq_t::arg_keywords to arg_keyword_num. * rename rb_iseq_t::arg_keyword to rb_iseq_t::arg_keyword_bits. to represent keyword bitmap parameter index. This bitmap parameter shows that which keyword parameters are given or not given (0 for given). It is refered by `checkkeyword' instruction described bellow. * rename rb_iseq_t::arg_keyword_check to rb_iseq_t::arg_keyword_rest to represent keyword rest parameter index. * add rb_iseq_t::arg_keyword_default_values to represent default keyword values. * rename VM_CALL_ARGS_SKIP_SETUP to VM_CALL_ARGS_SIMPLE to represent (ci->flag & (SPLAT|BLOCKARG)) && ci->blockiseq == NULL && ci->kw_arg == NULL. * vm_insnhelper.c, vm_args.c: rewrite with refactoring. * rewrite splat argument code. * rewrite keyword arguments/parameters code. * merge method and block parameter fitting code into one code base. * vm.c, vm_eval.c: catch up these changes. * compile.c (new_callinfo): callinfo requires kw_arg parameter. * compile.c (compile_array_): check the last argument Hash object or not. If Hash object and all keys are Symbol literals, they are compiled to keyword arguments. * insns.def (checkkeyword): add new instruction. This instruction check the availability of corresponding keyword. For example, a method "def foo k1: 'v1'; end" is cimpiled to the following instructions. 0000 checkkeyword 2, 0 # check k1 is given. 0003 branchif 9 # if given, jump to address #9 0005 putstring "v1" 0007 setlocal_OP__WC__0 3 # k1 = 'v1' 0009 trace 8 0011 putnil 0012 trace 16 0014 leave * insns.def (opt_send_simple): removed and add new instruction "opt_send_without_block". * parse.y (new_args_tail_gen): reorder variables. Before this patch, a method "def foo(k1: 1, kr1:, k2: 2, **krest, &b)" has parameter variables "k1, kr1, k2, &b, internal_id, krest", but this patch reorders to "kr1, k1, k2, internal_id, krest, &b". (locate a block variable at last) * parse.y (vtable_pop): added. This function remove latest `n' variables from vtable. * iseq.c: catch up iseq data changes. * proc.c: ditto. * class.c (keyword_error): export as rb_keyword_error(). * common.mk: depend vm_args.c for vm.o. * hash.c (rb_hash_has_key): export. * internal.h: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@48239 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2014-11-02 21:02:55 +03:00
if (ci->flag & VM_CALL_ARGS_SIMPLE) {
switch (ci->orig_argc) {
case 0:
switch (ci->mid) {
case idLength: SP_INSN(length); return COMPILE_OK;
case idSize: SP_INSN(size); return COMPILE_OK;
case idEmptyP: SP_INSN(empty_p);return COMPILE_OK;
case idNilP: SP_INSN(nil_p); return COMPILE_OK;
case idSucc: SP_INSN(succ); return COMPILE_OK;
case idNot: SP_INSN(not); return COMPILE_OK;
}
break;
case 1:
switch (ci->mid) {
case idPLUS: SP_INSN(plus); return COMPILE_OK;
case idMINUS: SP_INSN(minus); return COMPILE_OK;
case idMULT: SP_INSN(mult); return COMPILE_OK;
case idDIV: SP_INSN(div); return COMPILE_OK;
case idMOD: SP_INSN(mod); return COMPILE_OK;
case idEq: SP_INSN(eq); return COMPILE_OK;
case idNeq: SP_INSN(neq); return COMPILE_OK;
case idEqTilde:SP_INSN(regexpmatch2);return COMPILE_OK;
case idLT: SP_INSN(lt); return COMPILE_OK;
case idLE: SP_INSN(le); return COMPILE_OK;
case idGT: SP_INSN(gt); return COMPILE_OK;
case idGE: SP_INSN(ge); return COMPILE_OK;
case idLTLT: SP_INSN(ltlt); return COMPILE_OK;
case idAREF: SP_INSN(aref); return COMPILE_OK;
case idAnd: SP_INSN(and); return COMPILE_OK;
case idOr: SP_INSN(or); return COMPILE_OK;
}
break;
case 2:
switch (ci->mid) {
case idASET: SP_INSN(aset); return COMPILE_OK;
}
break;
}
}
* rewrite method/block parameter fitting logic to optimize keyword arguments/parameters and a splat argument. [Feature #10440] (Details are described in this ticket) Most of complex part is moved to vm_args.c. Now, ISeq#to_a does not catch up new instruction format. * vm_core.h: change iseq data structures. * introduce rb_call_info_kw_arg_t to represent keyword arguments. * add rb_call_info_t::kw_arg. * rename rb_iseq_t::arg_post_len to rb_iseq_t::arg_post_num. * rename rb_iseq_t::arg_keywords to arg_keyword_num. * rename rb_iseq_t::arg_keyword to rb_iseq_t::arg_keyword_bits. to represent keyword bitmap parameter index. This bitmap parameter shows that which keyword parameters are given or not given (0 for given). It is refered by `checkkeyword' instruction described bellow. * rename rb_iseq_t::arg_keyword_check to rb_iseq_t::arg_keyword_rest to represent keyword rest parameter index. * add rb_iseq_t::arg_keyword_default_values to represent default keyword values. * rename VM_CALL_ARGS_SKIP_SETUP to VM_CALL_ARGS_SIMPLE to represent (ci->flag & (SPLAT|BLOCKARG)) && ci->blockiseq == NULL && ci->kw_arg == NULL. * vm_insnhelper.c, vm_args.c: rewrite with refactoring. * rewrite splat argument code. * rewrite keyword arguments/parameters code. * merge method and block parameter fitting code into one code base. * vm.c, vm_eval.c: catch up these changes. * compile.c (new_callinfo): callinfo requires kw_arg parameter. * compile.c (compile_array_): check the last argument Hash object or not. If Hash object and all keys are Symbol literals, they are compiled to keyword arguments. * insns.def (checkkeyword): add new instruction. This instruction check the availability of corresponding keyword. For example, a method "def foo k1: 'v1'; end" is cimpiled to the following instructions. 0000 checkkeyword 2, 0 # check k1 is given. 0003 branchif 9 # if given, jump to address #9 0005 putstring "v1" 0007 setlocal_OP__WC__0 3 # k1 = 'v1' 0009 trace 8 0011 putnil 0012 trace 16 0014 leave * insns.def (opt_send_simple): removed and add new instruction "opt_send_without_block". * parse.y (new_args_tail_gen): reorder variables. Before this patch, a method "def foo(k1: 1, kr1:, k2: 2, **krest, &b)" has parameter variables "k1, kr1, k2, &b, internal_id, krest", but this patch reorders to "kr1, k1, k2, internal_id, krest, &b". (locate a block variable at last) * parse.y (vtable_pop): added. This function remove latest `n' variables from vtable. * iseq.c: catch up iseq data changes. * proc.c: ditto. * class.c (keyword_error): export as rb_keyword_error(). * common.mk: depend vm_args.c for vm.o. * hash.c (rb_hash_has_key): export. * internal.h: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@48239 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2014-11-02 21:02:55 +03:00
if ((ci->flag & VM_CALL_ARGS_BLOCKARG) == 0 && blockiseq == NULL) {
* rewrite method/block parameter fitting logic to optimize keyword arguments/parameters and a splat argument. [Feature #10440] (Details are described in this ticket) Most of complex part is moved to vm_args.c. Now, ISeq#to_a does not catch up new instruction format. * vm_core.h: change iseq data structures. * introduce rb_call_info_kw_arg_t to represent keyword arguments. * add rb_call_info_t::kw_arg. * rename rb_iseq_t::arg_post_len to rb_iseq_t::arg_post_num. * rename rb_iseq_t::arg_keywords to arg_keyword_num. * rename rb_iseq_t::arg_keyword to rb_iseq_t::arg_keyword_bits. to represent keyword bitmap parameter index. This bitmap parameter shows that which keyword parameters are given or not given (0 for given). It is refered by `checkkeyword' instruction described bellow. * rename rb_iseq_t::arg_keyword_check to rb_iseq_t::arg_keyword_rest to represent keyword rest parameter index. * add rb_iseq_t::arg_keyword_default_values to represent default keyword values. * rename VM_CALL_ARGS_SKIP_SETUP to VM_CALL_ARGS_SIMPLE to represent (ci->flag & (SPLAT|BLOCKARG)) && ci->blockiseq == NULL && ci->kw_arg == NULL. * vm_insnhelper.c, vm_args.c: rewrite with refactoring. * rewrite splat argument code. * rewrite keyword arguments/parameters code. * merge method and block parameter fitting code into one code base. * vm.c, vm_eval.c: catch up these changes. * compile.c (new_callinfo): callinfo requires kw_arg parameter. * compile.c (compile_array_): check the last argument Hash object or not. If Hash object and all keys are Symbol literals, they are compiled to keyword arguments. * insns.def (checkkeyword): add new instruction. This instruction check the availability of corresponding keyword. For example, a method "def foo k1: 'v1'; end" is cimpiled to the following instructions. 0000 checkkeyword 2, 0 # check k1 is given. 0003 branchif 9 # if given, jump to address #9 0005 putstring "v1" 0007 setlocal_OP__WC__0 3 # k1 = 'v1' 0009 trace 8 0011 putnil 0012 trace 16 0014 leave * insns.def (opt_send_simple): removed and add new instruction "opt_send_without_block". * parse.y (new_args_tail_gen): reorder variables. Before this patch, a method "def foo(k1: 1, kr1:, k2: 2, **krest, &b)" has parameter variables "k1, kr1, k2, &b, internal_id, krest", but this patch reorders to "kr1, k1, k2, internal_id, krest, &b". (locate a block variable at last) * parse.y (vtable_pop): added. This function remove latest `n' variables from vtable. * iseq.c: catch up iseq data changes. * proc.c: ditto. * class.c (keyword_error): export as rb_keyword_error(). * common.mk: depend vm_args.c for vm.o. * hash.c (rb_hash_has_key): export. * internal.h: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@48239 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2014-11-02 21:02:55 +03:00
iobj->insn_id = BIN(opt_send_without_block);
iobj->operand_size = insn_len(iobj->insn_id) - 1;
}
}
#undef SP_INSN
return COMPILE_OK;
}
static inline int
tailcallable_p(rb_iseq_t *iseq)
{
switch (iseq->body->type) {
case ISEQ_TYPE_TOP:
case ISEQ_TYPE_EVAL:
case ISEQ_TYPE_MAIN:
/* not tail callable because cfp will be over popped */
case ISEQ_TYPE_RESCUE:
case ISEQ_TYPE_ENSURE:
/* rescue block can't tail call because of errinfo */
return FALSE;
default:
return TRUE;
}
}
static int
iseq_optimize(rb_iseq_t *iseq, LINK_ANCHOR *const anchor)
{
LINK_ELEMENT *list;
const int do_peepholeopt = ISEQ_COMPILE_DATA(iseq)->option->peephole_optimization;
const int do_tailcallopt = tailcallable_p(iseq) &&
ISEQ_COMPILE_DATA(iseq)->option->tailcall_optimization;
const int do_si = ISEQ_COMPILE_DATA(iseq)->option->specialized_instruction;
const int do_ou = ISEQ_COMPILE_DATA(iseq)->option->operands_unification;
int rescue_level = 0;
int tailcallopt = do_tailcallopt;
list = FIRST_ELEMENT(anchor);
while (list) {
if (IS_INSN(list)) {
if (do_peepholeopt) {
iseq_peephole_optimize(iseq, list, tailcallopt);
}
if (do_si) {
iseq_specialized_instruction(iseq, (INSN *)list);
}
if (do_ou) {
insn_operands_unification((INSN *)list);
}
}
if (IS_LABEL(list)) {
switch (((LABEL *)list)->rescued) {
case LABEL_RESCUE_BEG:
rescue_level++;
tailcallopt = FALSE;
break;
case LABEL_RESCUE_END:
if (!--rescue_level) tailcallopt = do_tailcallopt;
break;
}
}
list = list->next;
}
return COMPILE_OK;
}
#if OPT_INSTRUCTIONS_UNIFICATION
static INSN *
new_unified_insn(rb_iseq_t *iseq,
int insn_id, int size, LINK_ELEMENT *seq_list)
{
INSN *iobj = 0;
LINK_ELEMENT *list = seq_list;
int i, argc = 0;
VALUE *operands = 0, *ptr = 0;
/* count argc */
for (i = 0; i < size; i++) {
iobj = (INSN *)list;
argc += iobj->operand_size;
list = list->next;
}
if (argc > 0) {
ptr = operands =
compile_data_alloc2(iseq, sizeof(VALUE), argc);
}
/* copy operands */
list = seq_list;
for (i = 0; i < size; i++) {
iobj = (INSN *)list;
MEMCPY(ptr, iobj->operands, VALUE, iobj->operand_size);
ptr += iobj->operand_size;
list = list->next;
}
return new_insn_core(iseq, iobj->insn_info.line_no, insn_id, argc, operands);
}
#endif
/*
* This scheme can get more performance if do this optimize with
* label address resolving.
* It's future work (if compile time was bottle neck).
*/
static int
iseq_insns_unification(rb_iseq_t *iseq, LINK_ANCHOR *const anchor)
{
#if OPT_INSTRUCTIONS_UNIFICATION
LINK_ELEMENT *list;
INSN *iobj, *niobj;
int id, k;
intptr_t j;
list = FIRST_ELEMENT(anchor);
while (list) {
if (IS_INSN(list)) {
iobj = (INSN *)list;
id = iobj->insn_id;
if (unified_insns_data[id] != 0) {
const int *const *entry = unified_insns_data[id];
for (j = 1; j < (intptr_t)entry[0]; j++) {
const int *unified = entry[j];
LINK_ELEMENT *li = list->next;
for (k = 2; k < unified[1]; k++) {
if (!IS_INSN(li) ||
((INSN *)li)->insn_id != unified[k]) {
goto miss;
}
li = li->next;
}
/* matched */
niobj =
new_unified_insn(iseq, unified[0], unified[1] - 1,
list);
/* insert to list */
niobj->link.prev = (LINK_ELEMENT *)iobj->link.prev;
niobj->link.next = li;
if (li) {
li->prev = (LINK_ELEMENT *)niobj;
}
list->prev->next = (LINK_ELEMENT *)niobj;
list = (LINK_ELEMENT *)niobj;
break;
miss:;
}
}
}
list = list->next;
}
#endif
return COMPILE_OK;
}
#if OPT_STACK_CACHING
#define SC_INSN(insn, stat) sc_insn_info[(insn)][(stat)]
#define SC_NEXT(insn) sc_insn_next[(insn)]
#include "opt_sc.inc"
static int
insn_set_sc_state(rb_iseq_t *iseq, const LINK_ELEMENT *anchor, INSN *iobj, int state)
{
int nstate;
int insn_id;
insn_id = iobj->insn_id;
iobj->insn_id = SC_INSN(insn_id, state);
nstate = SC_NEXT(iobj->insn_id);
if (insn_id == BIN(jump) ||
insn_id == BIN(branchif) || insn_id == BIN(branchunless)) {
LABEL *lobj = (LABEL *)OPERAND_AT(iobj, 0);
if (lobj->sc_state != 0) {
if (lobj->sc_state != nstate) {
BADINSN_DUMP(anchor, iobj, lobj);
COMPILE_ERROR(iseq, iobj->insn_info.line_no,
"insn_set_sc_state error: %d at "LABEL_FORMAT
", %d expected\n",
lobj->sc_state, lobj->label_no, nstate);
return COMPILE_NG;
}
}
else {
lobj->sc_state = nstate;
}
if (insn_id == BIN(jump)) {
nstate = SCS_XX;
}
}
else if (insn_id == BIN(leave)) {
nstate = SCS_XX;
}
return nstate;
}
static int
label_set_sc_state(LABEL *lobj, int state)
{
if (lobj->sc_state != 0) {
if (lobj->sc_state != state) {
state = lobj->sc_state;
}
}
else {
lobj->sc_state = state;
}
return state;
}
#endif
static int
iseq_set_sequence_stackcaching(rb_iseq_t *iseq, LINK_ANCHOR *const anchor)
{
#if OPT_STACK_CACHING
LINK_ELEMENT *list;
int state, insn_id;
/* initialize */
state = SCS_XX;
list = FIRST_ELEMENT(anchor);
/* dump_disasm_list(list); */
/* for each list element */
while (list) {
redo_point:
switch (list->type) {
case ISEQ_ELEMENT_INSN:
{
INSN *iobj = (INSN *)list;
insn_id = iobj->insn_id;
/* dump_disasm_list(list); */
switch (insn_id) {
case BIN(nop):
{
/* exception merge point */
if (state != SCS_AX) {
INSN *rpobj =
new_insn_body(iseq, 0, BIN(reput), 0);
/* replace this insn */
ELEM_REPLACE(list, (LINK_ELEMENT *)rpobj);
list = (LINK_ELEMENT *)rpobj;
goto redo_point;
}
break;
}
case BIN(swap):
{
if (state == SCS_AB || state == SCS_BA) {
state = (state == SCS_AB ? SCS_BA : SCS_AB);
ELEM_REMOVE(list);
list = list->next;
goto redo_point;
}
break;
}
case BIN(pop):
{
switch (state) {
case SCS_AX:
case SCS_BX:
state = SCS_XX;
break;
case SCS_AB:
state = SCS_AX;
break;
case SCS_BA:
state = SCS_BX;
break;
case SCS_XX:
goto normal_insn;
default:
COMPILE_ERROR(iseq, iobj->insn_info.line_no,
"unreachable");
return COMPILE_NG;
}
/* remove useless pop */
ELEM_REMOVE(list);
list = list->next;
goto redo_point;
}
default:;
/* none */
} /* end of switch */
normal_insn:
state = insn_set_sc_state(iseq, anchor, iobj, state);
break;
}
case ISEQ_ELEMENT_LABEL:
{
LABEL *lobj;
lobj = (LABEL *)list;
state = label_set_sc_state(lobj, state);
}
default:
break;
}
list = list->next;
}
#endif
return COMPILE_OK;
}
static int
all_string_result_p(const NODE *node)
{
if (!node) return FALSE;
switch (nd_type(node)) {
case NODE_STR: case NODE_DSTR:
return TRUE;
case NODE_IF: case NODE_UNLESS:
if (!node->nd_body || !node->nd_else) return FALSE;
if (all_string_result_p(node->nd_body))
return all_string_result_p(node->nd_else);
return FALSE;
case NODE_AND: case NODE_OR:
if (!node->nd_2nd)
return all_string_result_p(node->nd_1st);
if (!all_string_result_p(node->nd_1st))
return FALSE;
return all_string_result_p(node->nd_2nd);
default:
return FALSE;
}
}
static int
compile_dstr_fragments(rb_iseq_t *iseq, LINK_ANCHOR *const ret, const NODE *const node, int *cntp)
{
const NODE *list = node->nd_next;
VALUE lit = node->nd_lit;
LINK_ELEMENT *first_lit = 0;
int cnt = 0;
debugp_param("nd_lit", lit);
if (!NIL_P(lit)) {
cnt++;
if (!RB_TYPE_P(lit, T_STRING)) {
COMPILE_ERROR(ERROR_ARGS "dstr: must be string: %s",
rb_builtin_type_name(TYPE(lit)));
return COMPILE_NG;
}
lit = rb_fstring(lit);
ADD_INSN1(ret, nd_line(node), putobject, lit);
RB_OBJ_WRITTEN(iseq, Qundef, lit);
if (RSTRING_LEN(lit) == 0) first_lit = LAST_ELEMENT(ret);
}
while (list) {
const NODE *const head = list->nd_head;
if (nd_type(head) == NODE_STR) {
lit = rb_fstring(head->nd_lit);
ADD_INSN1(ret, nd_line(head), putobject, lit);
RB_OBJ_WRITTEN(iseq, Qundef, lit);
lit = Qnil;
}
else {
CHECK(COMPILE(ret, "each string", head));
}
cnt++;
list = list->nd_next;
}
if (NIL_P(lit) && first_lit) {
ELEM_REMOVE(first_lit);
--cnt;
}
*cntp = cnt;
return COMPILE_OK;
}
static int
compile_dstr(rb_iseq_t *iseq, LINK_ANCHOR *const ret, const NODE *const node)
{
int cnt;
CHECK(compile_dstr_fragments(iseq, ret, node, &cnt));
ADD_INSN1(ret, nd_line(node), concatstrings, INT2FIX(cnt));
return COMPILE_OK;
}
static int
compile_dregx(rb_iseq_t *iseq, LINK_ANCHOR *const ret, const NODE *const node)
{
int cnt;
CHECK(compile_dstr_fragments(iseq, ret, node, &cnt));
ADD_INSN2(ret, nd_line(node), toregexp, INT2FIX(node->nd_cflag), INT2FIX(cnt));
return COMPILE_OK;
}
static int
compile_flip_flop(rb_iseq_t *iseq, LINK_ANCHOR *const ret, const NODE *const node, int again,
LABEL *then_label, LABEL *else_label)
{
const int line = nd_line(node);
LABEL *lend = NEW_LABEL(line);
rb_num_t cnt = ISEQ_FLIP_CNT_INCREMENT(iseq->body->local_iseq)
+ VM_SVAR_FLIPFLOP_START;
VALUE key = INT2FIX(cnt);
ADD_INSN2(ret, line, getspecial, key, INT2FIX(0));
ADD_INSNL(ret, line, branchif, lend);
/* *flip == 0 */
CHECK(COMPILE(ret, "flip2 beg", node->nd_beg));
ADD_INSNL(ret, line, branchunless, else_label);
ADD_INSN1(ret, line, putobject, Qtrue);
ADD_INSN1(ret, line, setspecial, key);
if (!again) {
ADD_INSNL(ret, line, jump, then_label);
}
/* *flip == 1 */
ADD_LABEL(ret, lend);
CHECK(COMPILE(ret, "flip2 end", node->nd_end));
ADD_INSNL(ret, line, branchunless, then_label);
ADD_INSN1(ret, line, putobject, Qfalse);
ADD_INSN1(ret, line, setspecial, key);
ADD_INSNL(ret, line, jump, then_label);
return COMPILE_OK;
}
static int
compile_branch_condition(rb_iseq_t *iseq, LINK_ANCHOR *const ret, const NODE *cond,
LABEL *then_label, LABEL *else_label)
{
again:
switch (nd_type(cond)) {
case NODE_AND:
{
LABEL *label = NEW_LABEL(nd_line(cond));
CHECK(compile_branch_condition(iseq, ret, cond->nd_1st, label,
else_label));
if (!label->refcnt) break;
ADD_LABEL(ret, label);
cond = cond->nd_2nd;
goto again;
}
case NODE_OR:
{
LABEL *label = NEW_LABEL(nd_line(cond));
CHECK(compile_branch_condition(iseq, ret, cond->nd_1st, then_label,
label));
if (!label->refcnt) break;
ADD_LABEL(ret, label);
cond = cond->nd_2nd;
goto again;
}
case NODE_LIT: /* NODE_LIT is always true */
case NODE_TRUE:
case NODE_STR:
case NODE_ZLIST:
case NODE_LAMBDA:
/* printf("useless condition eliminate (%s)\n", ruby_node_name(nd_type(cond))); */
ADD_INSNL(ret, nd_line(cond), jump, then_label);
break;
case NODE_FALSE:
case NODE_NIL:
/* printf("useless condition eliminate (%s)\n", ruby_node_name(nd_type(cond))); */
ADD_INSNL(ret, nd_line(cond), jump, else_label);
break;
case NODE_LIST:
case NODE_ARGSCAT:
case NODE_DREGX:
case NODE_DSTR:
CHECK(COMPILE_POPPED(ret, "branch condition", cond));
ADD_INSNL(ret, nd_line(cond), jump, then_label);
break;
case NODE_FLIP2:
CHECK(compile_flip_flop(iseq, ret, cond, TRUE, then_label, else_label));
break;
case NODE_FLIP3:
CHECK(compile_flip_flop(iseq, ret, cond, FALSE, then_label, else_label));
break;
case NODE_DEFINED:
CHECK(compile_defined_expr(iseq, ret, cond, Qfalse));
goto branch;
default:
CHECK(COMPILE(ret, "branch condition", cond));
branch:
ADD_INSNL(ret, nd_line(cond), branchunless, else_label);
ADD_INSNL(ret, nd_line(cond), jump, then_label);
break;
}
return COMPILE_OK;
}
static int
keyword_node_p(const NODE *const node)
{
return nd_type(node) == NODE_HASH && node->nd_brace == FALSE;
}
* rewrite method/block parameter fitting logic to optimize keyword arguments/parameters and a splat argument. [Feature #10440] (Details are described in this ticket) Most of complex part is moved to vm_args.c. Now, ISeq#to_a does not catch up new instruction format. * vm_core.h: change iseq data structures. * introduce rb_call_info_kw_arg_t to represent keyword arguments. * add rb_call_info_t::kw_arg. * rename rb_iseq_t::arg_post_len to rb_iseq_t::arg_post_num. * rename rb_iseq_t::arg_keywords to arg_keyword_num. * rename rb_iseq_t::arg_keyword to rb_iseq_t::arg_keyword_bits. to represent keyword bitmap parameter index. This bitmap parameter shows that which keyword parameters are given or not given (0 for given). It is refered by `checkkeyword' instruction described bellow. * rename rb_iseq_t::arg_keyword_check to rb_iseq_t::arg_keyword_rest to represent keyword rest parameter index. * add rb_iseq_t::arg_keyword_default_values to represent default keyword values. * rename VM_CALL_ARGS_SKIP_SETUP to VM_CALL_ARGS_SIMPLE to represent (ci->flag & (SPLAT|BLOCKARG)) && ci->blockiseq == NULL && ci->kw_arg == NULL. * vm_insnhelper.c, vm_args.c: rewrite with refactoring. * rewrite splat argument code. * rewrite keyword arguments/parameters code. * merge method and block parameter fitting code into one code base. * vm.c, vm_eval.c: catch up these changes. * compile.c (new_callinfo): callinfo requires kw_arg parameter. * compile.c (compile_array_): check the last argument Hash object or not. If Hash object and all keys are Symbol literals, they are compiled to keyword arguments. * insns.def (checkkeyword): add new instruction. This instruction check the availability of corresponding keyword. For example, a method "def foo k1: 'v1'; end" is cimpiled to the following instructions. 0000 checkkeyword 2, 0 # check k1 is given. 0003 branchif 9 # if given, jump to address #9 0005 putstring "v1" 0007 setlocal_OP__WC__0 3 # k1 = 'v1' 0009 trace 8 0011 putnil 0012 trace 16 0014 leave * insns.def (opt_send_simple): removed and add new instruction "opt_send_without_block". * parse.y (new_args_tail_gen): reorder variables. Before this patch, a method "def foo(k1: 1, kr1:, k2: 2, **krest, &b)" has parameter variables "k1, kr1, k2, &b, internal_id, krest", but this patch reorders to "kr1, k1, k2, internal_id, krest, &b". (locate a block variable at last) * parse.y (vtable_pop): added. This function remove latest `n' variables from vtable. * iseq.c: catch up iseq data changes. * proc.c: ditto. * class.c (keyword_error): export as rb_keyword_error(). * common.mk: depend vm_args.c for vm.o. * hash.c (rb_hash_has_key): export. * internal.h: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@48239 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2014-11-02 21:02:55 +03:00
static int
compile_keyword_arg(rb_iseq_t *iseq, LINK_ANCHOR *const ret,
const NODE *const root_node,
struct rb_call_info_kw_arg **const kw_arg_ptr,
unsigned int *flag)
* rewrite method/block parameter fitting logic to optimize keyword arguments/parameters and a splat argument. [Feature #10440] (Details are described in this ticket) Most of complex part is moved to vm_args.c. Now, ISeq#to_a does not catch up new instruction format. * vm_core.h: change iseq data structures. * introduce rb_call_info_kw_arg_t to represent keyword arguments. * add rb_call_info_t::kw_arg. * rename rb_iseq_t::arg_post_len to rb_iseq_t::arg_post_num. * rename rb_iseq_t::arg_keywords to arg_keyword_num. * rename rb_iseq_t::arg_keyword to rb_iseq_t::arg_keyword_bits. to represent keyword bitmap parameter index. This bitmap parameter shows that which keyword parameters are given or not given (0 for given). It is refered by `checkkeyword' instruction described bellow. * rename rb_iseq_t::arg_keyword_check to rb_iseq_t::arg_keyword_rest to represent keyword rest parameter index. * add rb_iseq_t::arg_keyword_default_values to represent default keyword values. * rename VM_CALL_ARGS_SKIP_SETUP to VM_CALL_ARGS_SIMPLE to represent (ci->flag & (SPLAT|BLOCKARG)) && ci->blockiseq == NULL && ci->kw_arg == NULL. * vm_insnhelper.c, vm_args.c: rewrite with refactoring. * rewrite splat argument code. * rewrite keyword arguments/parameters code. * merge method and block parameter fitting code into one code base. * vm.c, vm_eval.c: catch up these changes. * compile.c (new_callinfo): callinfo requires kw_arg parameter. * compile.c (compile_array_): check the last argument Hash object or not. If Hash object and all keys are Symbol literals, they are compiled to keyword arguments. * insns.def (checkkeyword): add new instruction. This instruction check the availability of corresponding keyword. For example, a method "def foo k1: 'v1'; end" is cimpiled to the following instructions. 0000 checkkeyword 2, 0 # check k1 is given. 0003 branchif 9 # if given, jump to address #9 0005 putstring "v1" 0007 setlocal_OP__WC__0 3 # k1 = 'v1' 0009 trace 8 0011 putnil 0012 trace 16 0014 leave * insns.def (opt_send_simple): removed and add new instruction "opt_send_without_block". * parse.y (new_args_tail_gen): reorder variables. Before this patch, a method "def foo(k1: 1, kr1:, k2: 2, **krest, &b)" has parameter variables "k1, kr1, k2, &b, internal_id, krest", but this patch reorders to "kr1, k1, k2, internal_id, krest, &b". (locate a block variable at last) * parse.y (vtable_pop): added. This function remove latest `n' variables from vtable. * iseq.c: catch up iseq data changes. * proc.c: ditto. * class.c (keyword_error): export as rb_keyword_error(). * common.mk: depend vm_args.c for vm.o. * hash.c (rb_hash_has_key): export. * internal.h: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@48239 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2014-11-02 21:02:55 +03:00
{
if (kw_arg_ptr == NULL) return FALSE;
if (keyword_node_p(root_node) && root_node->nd_head && nd_type(root_node->nd_head) == NODE_LIST) {
const NODE *node = root_node->nd_head;
* rewrite method/block parameter fitting logic to optimize keyword arguments/parameters and a splat argument. [Feature #10440] (Details are described in this ticket) Most of complex part is moved to vm_args.c. Now, ISeq#to_a does not catch up new instruction format. * vm_core.h: change iseq data structures. * introduce rb_call_info_kw_arg_t to represent keyword arguments. * add rb_call_info_t::kw_arg. * rename rb_iseq_t::arg_post_len to rb_iseq_t::arg_post_num. * rename rb_iseq_t::arg_keywords to arg_keyword_num. * rename rb_iseq_t::arg_keyword to rb_iseq_t::arg_keyword_bits. to represent keyword bitmap parameter index. This bitmap parameter shows that which keyword parameters are given or not given (0 for given). It is refered by `checkkeyword' instruction described bellow. * rename rb_iseq_t::arg_keyword_check to rb_iseq_t::arg_keyword_rest to represent keyword rest parameter index. * add rb_iseq_t::arg_keyword_default_values to represent default keyword values. * rename VM_CALL_ARGS_SKIP_SETUP to VM_CALL_ARGS_SIMPLE to represent (ci->flag & (SPLAT|BLOCKARG)) && ci->blockiseq == NULL && ci->kw_arg == NULL. * vm_insnhelper.c, vm_args.c: rewrite with refactoring. * rewrite splat argument code. * rewrite keyword arguments/parameters code. * merge method and block parameter fitting code into one code base. * vm.c, vm_eval.c: catch up these changes. * compile.c (new_callinfo): callinfo requires kw_arg parameter. * compile.c (compile_array_): check the last argument Hash object or not. If Hash object and all keys are Symbol literals, they are compiled to keyword arguments. * insns.def (checkkeyword): add new instruction. This instruction check the availability of corresponding keyword. For example, a method "def foo k1: 'v1'; end" is cimpiled to the following instructions. 0000 checkkeyword 2, 0 # check k1 is given. 0003 branchif 9 # if given, jump to address #9 0005 putstring "v1" 0007 setlocal_OP__WC__0 3 # k1 = 'v1' 0009 trace 8 0011 putnil 0012 trace 16 0014 leave * insns.def (opt_send_simple): removed and add new instruction "opt_send_without_block". * parse.y (new_args_tail_gen): reorder variables. Before this patch, a method "def foo(k1: 1, kr1:, k2: 2, **krest, &b)" has parameter variables "k1, kr1, k2, &b, internal_id, krest", but this patch reorders to "kr1, k1, k2, internal_id, krest, &b". (locate a block variable at last) * parse.y (vtable_pop): added. This function remove latest `n' variables from vtable. * iseq.c: catch up iseq data changes. * proc.c: ditto. * class.c (keyword_error): export as rb_keyword_error(). * common.mk: depend vm_args.c for vm.o. * hash.c (rb_hash_has_key): export. * internal.h: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@48239 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2014-11-02 21:02:55 +03:00
while (node) {
const NODE *key_node = node->nd_head;
* rewrite method/block parameter fitting logic to optimize keyword arguments/parameters and a splat argument. [Feature #10440] (Details are described in this ticket) Most of complex part is moved to vm_args.c. Now, ISeq#to_a does not catch up new instruction format. * vm_core.h: change iseq data structures. * introduce rb_call_info_kw_arg_t to represent keyword arguments. * add rb_call_info_t::kw_arg. * rename rb_iseq_t::arg_post_len to rb_iseq_t::arg_post_num. * rename rb_iseq_t::arg_keywords to arg_keyword_num. * rename rb_iseq_t::arg_keyword to rb_iseq_t::arg_keyword_bits. to represent keyword bitmap parameter index. This bitmap parameter shows that which keyword parameters are given or not given (0 for given). It is refered by `checkkeyword' instruction described bellow. * rename rb_iseq_t::arg_keyword_check to rb_iseq_t::arg_keyword_rest to represent keyword rest parameter index. * add rb_iseq_t::arg_keyword_default_values to represent default keyword values. * rename VM_CALL_ARGS_SKIP_SETUP to VM_CALL_ARGS_SIMPLE to represent (ci->flag & (SPLAT|BLOCKARG)) && ci->blockiseq == NULL && ci->kw_arg == NULL. * vm_insnhelper.c, vm_args.c: rewrite with refactoring. * rewrite splat argument code. * rewrite keyword arguments/parameters code. * merge method and block parameter fitting code into one code base. * vm.c, vm_eval.c: catch up these changes. * compile.c (new_callinfo): callinfo requires kw_arg parameter. * compile.c (compile_array_): check the last argument Hash object or not. If Hash object and all keys are Symbol literals, they are compiled to keyword arguments. * insns.def (checkkeyword): add new instruction. This instruction check the availability of corresponding keyword. For example, a method "def foo k1: 'v1'; end" is cimpiled to the following instructions. 0000 checkkeyword 2, 0 # check k1 is given. 0003 branchif 9 # if given, jump to address #9 0005 putstring "v1" 0007 setlocal_OP__WC__0 3 # k1 = 'v1' 0009 trace 8 0011 putnil 0012 trace 16 0014 leave * insns.def (opt_send_simple): removed and add new instruction "opt_send_without_block". * parse.y (new_args_tail_gen): reorder variables. Before this patch, a method "def foo(k1: 1, kr1:, k2: 2, **krest, &b)" has parameter variables "k1, kr1, k2, &b, internal_id, krest", but this patch reorders to "kr1, k1, k2, internal_id, krest, &b". (locate a block variable at last) * parse.y (vtable_pop): added. This function remove latest `n' variables from vtable. * iseq.c: catch up iseq data changes. * proc.c: ditto. * class.c (keyword_error): export as rb_keyword_error(). * common.mk: depend vm_args.c for vm.o. * hash.c (rb_hash_has_key): export. * internal.h: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@48239 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2014-11-02 21:02:55 +03:00
assert(nd_type(node) == NODE_LIST);
if (!key_node) {
if (flag) *flag |= VM_CALL_KW_SPLAT;
return FALSE;
}
else if (nd_type(key_node) == NODE_LIT && RB_TYPE_P(key_node->nd_lit, T_SYMBOL)) {
* rewrite method/block parameter fitting logic to optimize keyword arguments/parameters and a splat argument. [Feature #10440] (Details are described in this ticket) Most of complex part is moved to vm_args.c. Now, ISeq#to_a does not catch up new instruction format. * vm_core.h: change iseq data structures. * introduce rb_call_info_kw_arg_t to represent keyword arguments. * add rb_call_info_t::kw_arg. * rename rb_iseq_t::arg_post_len to rb_iseq_t::arg_post_num. * rename rb_iseq_t::arg_keywords to arg_keyword_num. * rename rb_iseq_t::arg_keyword to rb_iseq_t::arg_keyword_bits. to represent keyword bitmap parameter index. This bitmap parameter shows that which keyword parameters are given or not given (0 for given). It is refered by `checkkeyword' instruction described bellow. * rename rb_iseq_t::arg_keyword_check to rb_iseq_t::arg_keyword_rest to represent keyword rest parameter index. * add rb_iseq_t::arg_keyword_default_values to represent default keyword values. * rename VM_CALL_ARGS_SKIP_SETUP to VM_CALL_ARGS_SIMPLE to represent (ci->flag & (SPLAT|BLOCKARG)) && ci->blockiseq == NULL && ci->kw_arg == NULL. * vm_insnhelper.c, vm_args.c: rewrite with refactoring. * rewrite splat argument code. * rewrite keyword arguments/parameters code. * merge method and block parameter fitting code into one code base. * vm.c, vm_eval.c: catch up these changes. * compile.c (new_callinfo): callinfo requires kw_arg parameter. * compile.c (compile_array_): check the last argument Hash object or not. If Hash object and all keys are Symbol literals, they are compiled to keyword arguments. * insns.def (checkkeyword): add new instruction. This instruction check the availability of corresponding keyword. For example, a method "def foo k1: 'v1'; end" is cimpiled to the following instructions. 0000 checkkeyword 2, 0 # check k1 is given. 0003 branchif 9 # if given, jump to address #9 0005 putstring "v1" 0007 setlocal_OP__WC__0 3 # k1 = 'v1' 0009 trace 8 0011 putnil 0012 trace 16 0014 leave * insns.def (opt_send_simple): removed and add new instruction "opt_send_without_block". * parse.y (new_args_tail_gen): reorder variables. Before this patch, a method "def foo(k1: 1, kr1:, k2: 2, **krest, &b)" has parameter variables "k1, kr1, k2, &b, internal_id, krest", but this patch reorders to "kr1, k1, k2, internal_id, krest, &b". (locate a block variable at last) * parse.y (vtable_pop): added. This function remove latest `n' variables from vtable. * iseq.c: catch up iseq data changes. * proc.c: ditto. * class.c (keyword_error): export as rb_keyword_error(). * common.mk: depend vm_args.c for vm.o. * hash.c (rb_hash_has_key): export. * internal.h: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@48239 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2014-11-02 21:02:55 +03:00
/* can be keywords */
}
else {
if (flag) *flag |= VM_CALL_KW_SPLAT;
* rewrite method/block parameter fitting logic to optimize keyword arguments/parameters and a splat argument. [Feature #10440] (Details are described in this ticket) Most of complex part is moved to vm_args.c. Now, ISeq#to_a does not catch up new instruction format. * vm_core.h: change iseq data structures. * introduce rb_call_info_kw_arg_t to represent keyword arguments. * add rb_call_info_t::kw_arg. * rename rb_iseq_t::arg_post_len to rb_iseq_t::arg_post_num. * rename rb_iseq_t::arg_keywords to arg_keyword_num. * rename rb_iseq_t::arg_keyword to rb_iseq_t::arg_keyword_bits. to represent keyword bitmap parameter index. This bitmap parameter shows that which keyword parameters are given or not given (0 for given). It is refered by `checkkeyword' instruction described bellow. * rename rb_iseq_t::arg_keyword_check to rb_iseq_t::arg_keyword_rest to represent keyword rest parameter index. * add rb_iseq_t::arg_keyword_default_values to represent default keyword values. * rename VM_CALL_ARGS_SKIP_SETUP to VM_CALL_ARGS_SIMPLE to represent (ci->flag & (SPLAT|BLOCKARG)) && ci->blockiseq == NULL && ci->kw_arg == NULL. * vm_insnhelper.c, vm_args.c: rewrite with refactoring. * rewrite splat argument code. * rewrite keyword arguments/parameters code. * merge method and block parameter fitting code into one code base. * vm.c, vm_eval.c: catch up these changes. * compile.c (new_callinfo): callinfo requires kw_arg parameter. * compile.c (compile_array_): check the last argument Hash object or not. If Hash object and all keys are Symbol literals, they are compiled to keyword arguments. * insns.def (checkkeyword): add new instruction. This instruction check the availability of corresponding keyword. For example, a method "def foo k1: 'v1'; end" is cimpiled to the following instructions. 0000 checkkeyword 2, 0 # check k1 is given. 0003 branchif 9 # if given, jump to address #9 0005 putstring "v1" 0007 setlocal_OP__WC__0 3 # k1 = 'v1' 0009 trace 8 0011 putnil 0012 trace 16 0014 leave * insns.def (opt_send_simple): removed and add new instruction "opt_send_without_block". * parse.y (new_args_tail_gen): reorder variables. Before this patch, a method "def foo(k1: 1, kr1:, k2: 2, **krest, &b)" has parameter variables "k1, kr1, k2, &b, internal_id, krest", but this patch reorders to "kr1, k1, k2, internal_id, krest, &b". (locate a block variable at last) * parse.y (vtable_pop): added. This function remove latest `n' variables from vtable. * iseq.c: catch up iseq data changes. * proc.c: ditto. * class.c (keyword_error): export as rb_keyword_error(). * common.mk: depend vm_args.c for vm.o. * hash.c (rb_hash_has_key): export. * internal.h: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@48239 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2014-11-02 21:02:55 +03:00
return FALSE;
}
node = node->nd_next; /* skip value node */
node = node->nd_next;
}
/* may be keywords */
node = root_node->nd_head;
{
int len = (int)node->nd_alen / 2;
struct rb_call_info_kw_arg *kw_arg =
rb_xmalloc_mul_add(len - 1, sizeof(VALUE), sizeof(struct rb_call_info_kw_arg));
VALUE *keywords = kw_arg->keywords;
* rewrite method/block parameter fitting logic to optimize keyword arguments/parameters and a splat argument. [Feature #10440] (Details are described in this ticket) Most of complex part is moved to vm_args.c. Now, ISeq#to_a does not catch up new instruction format. * vm_core.h: change iseq data structures. * introduce rb_call_info_kw_arg_t to represent keyword arguments. * add rb_call_info_t::kw_arg. * rename rb_iseq_t::arg_post_len to rb_iseq_t::arg_post_num. * rename rb_iseq_t::arg_keywords to arg_keyword_num. * rename rb_iseq_t::arg_keyword to rb_iseq_t::arg_keyword_bits. to represent keyword bitmap parameter index. This bitmap parameter shows that which keyword parameters are given or not given (0 for given). It is refered by `checkkeyword' instruction described bellow. * rename rb_iseq_t::arg_keyword_check to rb_iseq_t::arg_keyword_rest to represent keyword rest parameter index. * add rb_iseq_t::arg_keyword_default_values to represent default keyword values. * rename VM_CALL_ARGS_SKIP_SETUP to VM_CALL_ARGS_SIMPLE to represent (ci->flag & (SPLAT|BLOCKARG)) && ci->blockiseq == NULL && ci->kw_arg == NULL. * vm_insnhelper.c, vm_args.c: rewrite with refactoring. * rewrite splat argument code. * rewrite keyword arguments/parameters code. * merge method and block parameter fitting code into one code base. * vm.c, vm_eval.c: catch up these changes. * compile.c (new_callinfo): callinfo requires kw_arg parameter. * compile.c (compile_array_): check the last argument Hash object or not. If Hash object and all keys are Symbol literals, they are compiled to keyword arguments. * insns.def (checkkeyword): add new instruction. This instruction check the availability of corresponding keyword. For example, a method "def foo k1: 'v1'; end" is cimpiled to the following instructions. 0000 checkkeyword 2, 0 # check k1 is given. 0003 branchif 9 # if given, jump to address #9 0005 putstring "v1" 0007 setlocal_OP__WC__0 3 # k1 = 'v1' 0009 trace 8 0011 putnil 0012 trace 16 0014 leave * insns.def (opt_send_simple): removed and add new instruction "opt_send_without_block". * parse.y (new_args_tail_gen): reorder variables. Before this patch, a method "def foo(k1: 1, kr1:, k2: 2, **krest, &b)" has parameter variables "k1, kr1, k2, &b, internal_id, krest", but this patch reorders to "kr1, k1, k2, internal_id, krest, &b". (locate a block variable at last) * parse.y (vtable_pop): added. This function remove latest `n' variables from vtable. * iseq.c: catch up iseq data changes. * proc.c: ditto. * class.c (keyword_error): export as rb_keyword_error(). * common.mk: depend vm_args.c for vm.o. * hash.c (rb_hash_has_key): export. * internal.h: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@48239 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2014-11-02 21:02:55 +03:00
int i = 0;
kw_arg->keyword_len = len;
*kw_arg_ptr = kw_arg;
for (i=0; node != NULL; i++, node = node->nd_next->nd_next) {
const NODE *key_node = node->nd_head;
const NODE *val_node = node->nd_next->nd_head;
keywords[i] = key_node->nd_lit;
2019-07-15 08:43:39 +03:00
NO_CHECK(COMPILE(ret, "keyword values", val_node));
* rewrite method/block parameter fitting logic to optimize keyword arguments/parameters and a splat argument. [Feature #10440] (Details are described in this ticket) Most of complex part is moved to vm_args.c. Now, ISeq#to_a does not catch up new instruction format. * vm_core.h: change iseq data structures. * introduce rb_call_info_kw_arg_t to represent keyword arguments. * add rb_call_info_t::kw_arg. * rename rb_iseq_t::arg_post_len to rb_iseq_t::arg_post_num. * rename rb_iseq_t::arg_keywords to arg_keyword_num. * rename rb_iseq_t::arg_keyword to rb_iseq_t::arg_keyword_bits. to represent keyword bitmap parameter index. This bitmap parameter shows that which keyword parameters are given or not given (0 for given). It is refered by `checkkeyword' instruction described bellow. * rename rb_iseq_t::arg_keyword_check to rb_iseq_t::arg_keyword_rest to represent keyword rest parameter index. * add rb_iseq_t::arg_keyword_default_values to represent default keyword values. * rename VM_CALL_ARGS_SKIP_SETUP to VM_CALL_ARGS_SIMPLE to represent (ci->flag & (SPLAT|BLOCKARG)) && ci->blockiseq == NULL && ci->kw_arg == NULL. * vm_insnhelper.c, vm_args.c: rewrite with refactoring. * rewrite splat argument code. * rewrite keyword arguments/parameters code. * merge method and block parameter fitting code into one code base. * vm.c, vm_eval.c: catch up these changes. * compile.c (new_callinfo): callinfo requires kw_arg parameter. * compile.c (compile_array_): check the last argument Hash object or not. If Hash object and all keys are Symbol literals, they are compiled to keyword arguments. * insns.def (checkkeyword): add new instruction. This instruction check the availability of corresponding keyword. For example, a method "def foo k1: 'v1'; end" is cimpiled to the following instructions. 0000 checkkeyword 2, 0 # check k1 is given. 0003 branchif 9 # if given, jump to address #9 0005 putstring "v1" 0007 setlocal_OP__WC__0 3 # k1 = 'v1' 0009 trace 8 0011 putnil 0012 trace 16 0014 leave * insns.def (opt_send_simple): removed and add new instruction "opt_send_without_block". * parse.y (new_args_tail_gen): reorder variables. Before this patch, a method "def foo(k1: 1, kr1:, k2: 2, **krest, &b)" has parameter variables "k1, kr1, k2, &b, internal_id, krest", but this patch reorders to "kr1, k1, k2, internal_id, krest, &b". (locate a block variable at last) * parse.y (vtable_pop): added. This function remove latest `n' variables from vtable. * iseq.c: catch up iseq data changes. * proc.c: ditto. * class.c (keyword_error): export as rb_keyword_error(). * common.mk: depend vm_args.c for vm.o. * hash.c (rb_hash_has_key): export. * internal.h: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@48239 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2014-11-02 21:02:55 +03:00
}
assert(i == len);
return TRUE;
}
}
return FALSE;
}
static int
compile_args(rb_iseq_t *iseq, LINK_ANCHOR *const ret, const NODE *node,
struct rb_call_info_kw_arg **keywords_ptr, unsigned int *flag)
{
int len = 0;
for (; node; len++, node = node->nd_next) {
if (CPDEBUG > 0) {
EXPECT_NODE("compile_args", node, NODE_LIST, -1);
}
if (node->nd_next == NULL /* last node */ &&
compile_keyword_arg(iseq, ret, node->nd_head, keywords_ptr, flag)) {
len--;
}
else {
NO_CHECK(COMPILE_(ret, "array element", node->nd_head, FALSE));
}
}
return len;
}
static inline int
static_literal_node_p(const NODE *node, const rb_iseq_t *iseq)
{
node = node->nd_head;
switch (nd_type(node)) {
case NODE_LIT:
case NODE_NIL:
case NODE_TRUE:
case NODE_FALSE:
return TRUE;
case NODE_STR:
return ISEQ_COMPILE_DATA(iseq)->option->frozen_string_literal;
default:
return FALSE;
}
}
static inline VALUE
static_literal_value(const NODE *node, rb_iseq_t *iseq)
{
node = node->nd_head;
switch (nd_type(node)) {
case NODE_NIL:
return Qnil;
case NODE_TRUE:
return Qtrue;
case NODE_FALSE:
return Qfalse;
case NODE_STR:
if (ISEQ_COMPILE_DATA(iseq)->option->debug_frozen_string_literal || RTEST(ruby_debug)) {
VALUE lit;
VALUE debug_info = rb_ary_new_from_args(2, rb_iseq_path(iseq), INT2FIX((int)nd_line(node)));
lit = rb_str_dup(node->nd_lit);
rb_ivar_set(lit, id_debug_created_info, rb_obj_freeze(debug_info));
return rb_str_freeze(lit);
}
else {
return rb_fstring(node->nd_lit);
}
default:
return node->nd_lit;
}
}
static int
compile_array(rb_iseq_t *iseq, LINK_ANCHOR *const ret, const NODE *node, int popped)
{
int line = (int)nd_line(node);
if (nd_type(node) == NODE_ZLIST) {
if (!popped) {
ADD_INSN1(ret, line, newarray, INT2FIX(0));
}
return 0;
}
EXPECT_NODE("compile_array", node, NODE_LIST, -1);
if (popped) {
for (; node; node = node->nd_next) {
NO_CHECK(COMPILE_(ret, "array element", node->nd_head, popped));
}
return 1;
}
/* Compilation of an array literal.
* The following code is essentially the same as:
*
* for (int count = 0; node; count++; node->nd_next) {
* compile(node->nd_head);
* }
* ADD_INSN(newarray, count);
*
* However, there are three points.
*
* - The code above causes stack overflow for a big string literal.
* The following limits the stack length up to max_stack_len.
*
* [x1,x2,...,x10000] =>
2019-09-07 16:22:30 +03:00
* push x1 ; push x2 ; ...; push x256; newarray 256;
* push x257; push x258; ...; push x512; newarray 256; concatarray;
* push x513; push x514; ...; push x768; newarray 256; concatarray;
* ...
*
* - Long subarray can be optimized by pre-allocating a hidden array.
*
* [1,2,3,...,100] =>
* duparray [1,2,3,...,100]
*
* [x, 1,2,3,...,100, z] =>
* push x; newarray 1;
* putobject [1,2,3,...,100] (<- hidden array); concatarray;
* push z; newarray 1; concatarray
*
* - If the last element is a keyword, newarraykwsplat should be emitted
* to check and remove empty keyword arguments hash from array.
* (Note: a keyword is NODE_HASH which is not static_literal_node_p.)
*
* [1,2,3,**kw] =>
* putobject 1; putobject 2; putobject 3; push kw; newarraykwsplat
*/
const int max_stack_len = 0x100;
const int min_tmp_ary_len = 0x40;
int stack_len = 0;
int first_chunk = 1;
/* Convert pushed elements to an array, and concatarray if needed */
#define FLUSH_CHUNK(newarrayinsn) \
if (stack_len) { \
ADD_INSN1(ret, line, newarrayinsn, INT2FIX(stack_len)); \
if (!first_chunk) ADD_INSN(ret, line, concatarray); \
first_chunk = stack_len = 0; \
}
while (node) {
int count = 1;
/* pre-allocation check (this branch can be omittable) */
if (static_literal_node_p(node, iseq)) {
/* count the elements that are optimizable */
const NODE *node_tmp = node->nd_next;
2019-09-27 04:20:56 +03:00
for (; node_tmp && static_literal_node_p(node_tmp, iseq); node_tmp = node_tmp->nd_next)
count++;
if ((first_chunk && stack_len == 0 && !node_tmp) || count >= min_tmp_ary_len) {
/* The literal contains only optimizable elements, or the subarray is long enough */
VALUE ary = rb_ary_tmp_new(count);
/* Create a hidden array */
for (; count; count--, node = node->nd_next)
rb_ary_push(ary, static_literal_value(node, iseq));
OBJ_FREEZE(ary);
/* Emit optimized code */
FLUSH_CHUNK(newarray);
if (first_chunk) {
ADD_INSN1(ret, line, duparray, ary);
first_chunk = 0;
}
else {
ADD_INSN1(ret, line, putobject, ary);
ADD_INSN(ret, line, concatarray);
}
RB_OBJ_WRITTEN(iseq, Qundef, ary);
}
}
/* Base case: Compile "count" elements */
for (; count; count--, node = node->nd_next) {
if (CPDEBUG > 0) {
EXPECT_NODE("compile_array", node, NODE_LIST, -1);
}
NO_CHECK(COMPILE_(ret, "array element", node->nd_head, 0));
stack_len++;
if (!node->nd_next && keyword_node_p(node->nd_head)) {
/* Reached the end, and the last element is a keyword */
FLUSH_CHUNK(newarraykwsplat);
return 1;
}
/* If there are many pushed elements, flush them to avoid stack overflow */
if (stack_len >= max_stack_len) FLUSH_CHUNK(newarray);
}
}
FLUSH_CHUNK(newarray);
#undef FLUSH_CHUNK
return 1;
}
static inline int
static_literal_node_pair_p(const NODE *node, const rb_iseq_t *iseq)
{
return node->nd_head && static_literal_node_p(node, iseq) && static_literal_node_p(node->nd_next, iseq);
}
static int
compile_hash(rb_iseq_t *iseq, LINK_ANCHOR *const ret, const NODE *node, int popped)
{
int line = (int)nd_line(node);
node = node->nd_head;
if (!node || nd_type(node) == NODE_ZLIST) {
if (!popped) {
ADD_INSN1(ret, line, newhash, INT2FIX(0));
}
return 0;
}
EXPECT_NODE("compile_hash", node, NODE_LIST, -1);
if (popped) {
for (; node; node = node->nd_next) {
NO_CHECK(COMPILE_(ret, "hash element", node->nd_head, popped));
}
return 1;
}
/* Compilation of a hash literal (or keyword arguments).
* This is very similar to compile_array, but there are some differences:
*
2019-09-09 14:06:00 +03:00
* - It contains key-value pairs. So we need to take every two elements.
* We can assume that the length is always even.
*
* - Merging is done by a method call (id_core_hash_merge_ptr).
* Sometimes we need to insert the receiver, so "anchor" is needed.
* In addition, a method call is much slower than concatarray.
* So it pays only when the subsequence is really long.
* (min_tmp_hash_len must be much larger than min_tmp_ary_len.)
*
* - We need to handle keyword splat: **kw.
* For **kw, the key part (node->nd_head) is NULL, and the value part
* (node->nd_next->nd_head) is "kw".
* The code is a bit difficult to avoid hash allocation for **{}.
*/
const int max_stack_len = 0x100;
const int min_tmp_hash_len = 0x800;
int stack_len = 0;
int first_chunk = 1;
DECL_ANCHOR(anchor);
INIT_ANCHOR(anchor);
/* Convert pushed elements to a hash, and merge if needed */
#define FLUSH_CHUNK() \
if (stack_len) { \
if (first_chunk) { \
APPEND_LIST(ret, anchor); \
ADD_INSN1(ret, line, newhash, INT2FIX(stack_len)); \
} \
else { \
ADD_INSN1(ret, line, putspecialobject, INT2FIX(VM_SPECIAL_OBJECT_VMCORE)); \
ADD_INSN(ret, line, swap); \
APPEND_LIST(ret, anchor); \
ADD_SEND(ret, line, id_core_hash_merge_ptr, INT2FIX(stack_len + 1)); \
} \
INIT_ANCHOR(anchor); \
first_chunk = stack_len = 0; \
}
while (node) {
int count = 1;
/* pre-allocation check (this branch can be omittable) */
if (static_literal_node_pair_p(node, iseq)) {
/* count the elements that are optimizable */
const NODE *node_tmp = node->nd_next->nd_next;
2019-09-27 04:20:56 +03:00
for (; node_tmp && static_literal_node_pair_p(node_tmp, iseq); node_tmp = node_tmp->nd_next->nd_next)
count++;
if ((first_chunk && stack_len == 0 && !node_tmp) || count >= min_tmp_hash_len) {
/* The literal contains only optimizable elements, or the subsequence is long enough */
VALUE ary = rb_ary_tmp_new(count);
/* Create a hidden hash */
for (; count; count--, node = node->nd_next->nd_next) {
VALUE elem[2];
elem[0] = static_literal_value(node, iseq);
elem[1] = static_literal_value(node->nd_next, iseq);
rb_ary_cat(ary, elem, 2);
}
VALUE hash = rb_hash_new_with_size(RARRAY_LEN(ary) / 2);
rb_hash_bulk_insert(RARRAY_LEN(ary), RARRAY_CONST_PTR_TRANSIENT(ary), hash);
hash = rb_obj_hide(hash);
OBJ_FREEZE(hash);
/* Emit optimized code */
FLUSH_CHUNK();
if (first_chunk) {
ADD_INSN1(ret, line, duphash, hash);
first_chunk = 0;
}
else {
ADD_INSN1(ret, line, putspecialobject, INT2FIX(VM_SPECIAL_OBJECT_VMCORE));
ADD_INSN(ret, line, swap);
ADD_INSN1(ret, line, putobject, hash);
ADD_SEND(ret, line, id_core_hash_merge_kwd, INT2FIX(2));
}
RB_OBJ_WRITTEN(iseq, Qundef, hash);
}
}
/* Base case: Compile "count" elements */
for (; count; count--, node = node->nd_next->nd_next) {
Speed up hash literals by duping This commit replaces the `newhashfromarray` instruction with a `duphash` instruction. Instead of allocating a new hash from an array stored in the Instruction Sequences, store a hash directly in the instruction sequences and dup it on execution. == Instruction sequence changes == ```ruby code = <<-eorby { "foo" => "bar", "baz" => "lol" } eorby insns = RubyVM::InstructionSequence.compile(code, __FILE__, nil, 0, frozen_string_literal: true) puts insns.disasm ``` On Ruby 2.5: ``` == disasm: #<ISeq:<compiled>@test.rb:0 (0,0)-(0,36)>==================== 0000 putobject "foo" 0002 putobject "bar" 0004 putobject "baz" 0006 putobject "lol" 0008 newhash 4 0010 leave ``` Ruby 2.6@r66174 3b6321083a2e3525da3b34d08a0b68bac094bd7f: ``` $ ./ruby test.rb == disasm: #<ISeq:<compiled>@test.rb:0 (0,0)-(0,36)> (catch: FALSE) 0000 newhashfromarray 2, ["foo", "bar", "baz", "lol"] 0003 leave ``` Ruby 2.6 + This commit: ``` $ ./ruby test.rb == disasm: #<ISeq:<compiled>@test.rb:0 (0,0)-(0,36)> (catch: FALSE) 0000 duphash {"foo"=>"bar", "baz"=>"lol"} 0002 leave ``` == Benchmark Results == Compared to 2.5.3: ``` $ make benchmark ITEM=hash_literal_small COMPARE_RUBY=/Users/aaron/.rbenv/versions/2.5.3/bin/ruby generating known_errors.inc known_errors.inc unchanged ./revision.h unchanged /Users/aaron/.rbenv/shims/ruby --disable=gems -rrubygems -I./benchmark/lib ./benchmark/benchmark-driver/exe/benchmark-driver \ --executables="compare-ruby::/Users/aaron/.rbenv/versions/2.5.3/bin/ruby -I.ext/common --disable-gem" \ --executables="built-ruby::./miniruby -I./lib -I. -I.ext/common -r./prelude --disable-gem" \ $(find ./benchmark -maxdepth 1 -name '*hash_literal_small*.yml' -o -name '*hash_literal_small*.rb' | sort) Calculating ------------------------------------- compare-ruby built-ruby hash_literal_small2 1.498 1.877 i/s - 1.000 times in 0.667581s 0.532656s hash_literal_small4 1.197 1.642 i/s - 1.000 times in 0.835375s 0.609160s hash_literal_small8 0.620 1.215 i/s - 1.000 times in 1.611638s 0.823090s Comparison: hash_literal_small2 built-ruby: 1.9 i/s compare-ruby: 1.5 i/s - 1.25x slower hash_literal_small4 built-ruby: 1.6 i/s compare-ruby: 1.2 i/s - 1.37x slower hash_literal_small8 built-ruby: 1.2 i/s compare-ruby: 0.6 i/s - 1.96x slower ``` Compared to r66255 ``` $ make benchmark ITEM=hash_literal_small COMPARE_RUBY=/Users/aaron/.rbenv/versions/ruby-trunk/bin/ruby generating known_errors.inc known_errors.inc unchanged ./revision.h unchanged /Users/aaron/.rbenv/shims/ruby --disable=gems -rrubygems -I./benchmark/lib ./benchmark/benchmark-driver/exe/benchmark-driver \ --executables="compare-ruby::/Users/aaron/.rbenv/versions/ruby-trunk/bin/ruby -I.ext/common --disable-gem" \ --executables="built-ruby::./miniruby -I./lib -I. -I.ext/common -r./prelude --disable-gem" \ $(find ./benchmark -maxdepth 1 -name '*hash_literal_small*.yml' -o -name '*hash_literal_small*.rb' | sort) Calculating ------------------------------------- compare-ruby built-ruby hash_literal_small2 1.567 1.831 i/s - 1.000 times in 0.638056s 0.546039s hash_literal_small4 1.298 1.652 i/s - 1.000 times in 0.770214s 0.605182s hash_literal_small8 0.873 1.216 i/s - 1.000 times in 1.145304s 0.822047s Comparison: hash_literal_small2 built-ruby: 1.8 i/s compare-ruby: 1.6 i/s - 1.17x slower hash_literal_small4 built-ruby: 1.7 i/s compare-ruby: 1.3 i/s - 1.27x slower hash_literal_small8 built-ruby: 1.2 i/s compare-ruby: 0.9 i/s - 1.39x slower ``` git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@66258 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2018-12-06 21:28:21 +03:00
if (CPDEBUG > 0) {
EXPECT_NODE("compile_hash", node, NODE_LIST, -1);
}
Make m(**{}) mean call without keywords Previously, **{} was removed by the parser: ``` $ ruby --dump=parse -e '{**{}}' @ NODE_SCOPE (line: 1, location: (1,0)-(1,6)) +- nd_tbl: (empty) +- nd_args: | (null node) +- nd_body: @ NODE_HASH (line: 1, location: (1,0)-(1,6))* +- nd_brace: 1 (hash literal) +- nd_head: (null node) ``` Since it was removed by the parser, the compiler did not know about it, and `m(**{})` was therefore treated as `m()`. This modifies the parser to not remove the `**{}`. A simple approach for this is fairly simple by just removing a few lines from the parser, but that would cause two hash allocations every time it was used. The approach taken here modifies both the parser and the compiler, and results in `**{}` not allocating any hashes in the usual case. The basic idea is we use a literal node in the parser containing a frozen empty hash literal. In the compiler, we recognize when that is used, and if it is the only keyword present, we just push it onto the VM stack (no creation of a new hash or merging of keywords). If it is the first keyword present, we push a new empty hash onto the VM stack, so that later keywords can merge into it. If it is not the first keyword present, we can ignore it, since the there is no reason to merge an empty hash into the existing hash. Example instructions for `m(**{})` Before (note ARGS_SIMPLE): ``` == disasm: #<ISeq:<main>@-e:1 (1,0)-(1,7)> (catch: FALSE) 0000 putself ( 1)[Li] 0001 opt_send_without_block <callinfo!mid:m, argc:0, FCALL|ARGS_SIMPLE>, <callcache> 0004 leave ``` After (note putobject and KW_SPLAT): ``` == disasm: #<ISeq:<main>@-e:1 (1,0)-(1,7)> (catch: FALSE) 0000 putself ( 1)[Li] 0001 putobject {} 0003 opt_send_without_block <callinfo!mid:m, argc:1, FCALL|KW_SPLAT>, <callcache> 0006 leave ``` Example instructions for `m(**h, **{})` Before and After (no change): ``` == disasm: #<ISeq:<main>@-e:1 (1,0)-(1,12)> (catch: FALSE) 0000 putself ( 1)[Li] 0001 putspecialobject 1 0003 newhash 0 0005 putself 0006 opt_send_without_block <callinfo!mid:h, argc:0, FCALL|VCALL|ARGS_SIMPLE>, <callcache> 0009 opt_send_without_block <callinfo!mid:core#hash_merge_kwd, argc:2, ARGS_SIMPLE>, <callcache> 0012 opt_send_without_block <callinfo!mid:m, argc:1, FCALL|KW_SPLAT>, <callcache> 0015 leave ``` Example instructions for `m(**{}, **h)` Before: ``` == disasm: #<ISeq:<main>@-e:1 (1,0)-(1,12)> (catch: FALSE) 0000 putself ( 1)[Li] 0001 putspecialobject 1 0003 newhash 0 0005 putself 0006 opt_send_without_block <callinfo!mid:h, argc:0, FCALL|VCALL|ARGS_SIMPLE>, <callcache> 0009 opt_send_without_block <callinfo!mid:core#hash_merge_kwd, argc:2, ARGS_SIMPLE>, <callcache> 0012 opt_send_without_block <callinfo!mid:m, argc:1, FCALL|KW_SPLAT>, <callcache> 0015 leave ``` After (basically the same except for the addition of swap): ``` == disasm: #<ISeq:<main>@-e:1 (1,0)-(1,12)> (catch: FALSE) 0000 putself ( 1)[Li] 0001 newhash 0 0003 putspecialobject 1 0005 swap 0006 putself 0007 opt_send_without_block <callinfo!mid:h, argc:0, FCALL|VCALL|ARGS_SIMPLE>, <callcache> 0010 opt_send_without_block <callinfo!mid:core#hash_merge_kwd, argc:2, ARGS_SIMPLE>, <callcache> 0013 opt_send_without_block <callinfo!mid:m, argc:1, FCALL|KW_SPLAT>, <callcache> 0016 leave ```
2019-09-04 20:54:12 +03:00
if (node->nd_head) {
/* Normal key-value pair */
NO_CHECK(COMPILE_(anchor, "hash key element", node->nd_head, 0));
NO_CHECK(COMPILE_(anchor, "hash value element", node->nd_next->nd_head, 0));
stack_len += 2;
/* If there are many pushed elements, flush them to avoid stack overflow */
if (stack_len >= max_stack_len) FLUSH_CHUNK();
}
else {
/* kwsplat case: foo(..., **kw, ...) */
FLUSH_CHUNK();
const NODE *kw = node->nd_next->nd_head;
int empty_kw = nd_type(kw) == NODE_LIT; /* foo( ..., **{}, ...) */
int first_kw = first_chunk && stack_len == 0; /* foo(1,2,3, **kw, ...) */
int last_kw = !node->nd_next->nd_next; /* foo( ..., **kw) */
int only_kw = last_kw && first_kw; /* foo(1,2,3, **kw) */
if (empty_kw) {
if (only_kw) {
/* **{} appears at the last, so it won't be modified.
* kw is a special NODE_LIT that contains a special empty hash,
* so this emits: putobject {}
*/
NO_CHECK(COMPILE(ret, "keyword splat", kw));
}
else if (first_kw) {
/* **{} appears at the first, so it may be modified.
* We need to create a fresh hash object.
*/
ADD_INSN1(ret, line, newhash, INT2FIX(0));
}
}
else {
/* This is not empty hash: **{k:1}.
* We need to clone the hash (if first), or merge the hash to
* the accumulated hash (if not first).
*/
ADD_INSN1(ret, line, putspecialobject, INT2FIX(VM_SPECIAL_OBJECT_VMCORE));
if (first_kw) ADD_INSN1(ret, line, newhash, INT2FIX(0));
else ADD_INSN(ret, line, swap);
NO_CHECK(COMPILE(ret, "keyword splat", kw));
ADD_SEND(ret, line, id_core_hash_merge_kwd, INT2FIX(2));
}
first_chunk = 0;
}
}
}
FLUSH_CHUNK();
#undef FLUSH_CHUNK
return 1;
}
VALUE
rb_node_case_when_optimizable_literal(const NODE *const node)
{
switch (nd_type(node)) {
case NODE_LIT: {
VALUE v = node->nd_lit;
double ival;
if (RB_TYPE_P(v, T_FLOAT) &&
modf(RFLOAT_VALUE(v), &ival) == 0.0) {
return FIXABLE(ival) ? LONG2FIX((long)ival) : rb_dbl2big(ival);
}
if (SYMBOL_P(v) || rb_obj_is_kind_of(v, rb_cNumeric)) {
return v;
}
break;
}
case NODE_NIL:
return Qnil;
case NODE_TRUE:
return Qtrue;
case NODE_FALSE:
return Qfalse;
case NODE_STR:
return rb_fstring(node->nd_lit);
}
return Qundef;
}
static int
when_vals(rb_iseq_t *iseq, LINK_ANCHOR *const cond_seq, const NODE *vals,
LABEL *l1, int only_special_literals, VALUE literals)
{
while (vals) {
const NODE *val = vals->nd_head;
2019-05-05 05:13:54 +03:00
VALUE lit = rb_node_case_when_optimizable_literal(val);
if (lit == Qundef) {
only_special_literals = 0;
}
2019-05-05 05:13:54 +03:00
else if (NIL_P(rb_hash_lookup(literals, lit))) {
rb_hash_aset(literals, lit, (VALUE)(l1) | 1);
}
ADD_INSN(cond_seq, nd_line(val), dup); /* dup target */
if (nd_type(val) == NODE_STR) {
debugp_param("nd_lit", val->nd_lit);
lit = rb_fstring(val->nd_lit);
ADD_INSN1(cond_seq, nd_line(val), putobject, lit);
RB_OBJ_WRITTEN(iseq, Qundef, lit);
}
else {
if (!COMPILE(cond_seq, "when cond", val)) return -1;
}
ADD_INSN1(cond_seq, nd_line(vals), checkmatch, INT2FIX(VM_CHECKMATCH_TYPE_CASE));
ADD_INSNL(cond_seq, nd_line(val), branchif, l1);
vals = vals->nd_next;
}
return only_special_literals;
}
static int
when_splat_vals(rb_iseq_t *iseq, LINK_ANCHOR *const cond_seq, const NODE *vals,
LABEL *l1, int only_special_literals, VALUE literals)
{
const int line = nd_line(vals);
switch (nd_type(vals)) {
case NODE_LIST:
if (when_vals(iseq, cond_seq, vals, l1, only_special_literals, literals) < 0)
return COMPILE_NG;
break;
case NODE_SPLAT:
ADD_INSN (cond_seq, line, dup);
CHECK(COMPILE(cond_seq, "when splat", vals->nd_head));
ADD_INSN1(cond_seq, line, splatarray, Qfalse);
ADD_INSN1(cond_seq, line, checkmatch, INT2FIX(VM_CHECKMATCH_TYPE_CASE | VM_CHECKMATCH_ARRAY));
ADD_INSNL(cond_seq, line, branchif, l1);
break;
case NODE_ARGSCAT:
CHECK(when_splat_vals(iseq, cond_seq, vals->nd_head, l1, only_special_literals, literals));
CHECK(when_splat_vals(iseq, cond_seq, vals->nd_body, l1, only_special_literals, literals));
break;
case NODE_ARGSPUSH:
CHECK(when_splat_vals(iseq, cond_seq, vals->nd_head, l1, only_special_literals, literals));
ADD_INSN (cond_seq, line, dup);
CHECK(COMPILE(cond_seq, "when argspush body", vals->nd_body));
ADD_INSN1(cond_seq, line, checkmatch, INT2FIX(VM_CHECKMATCH_TYPE_CASE));
ADD_INSNL(cond_seq, line, branchif, l1);
break;
default:
ADD_INSN (cond_seq, line, dup);
CHECK(COMPILE(cond_seq, "when val", vals));
ADD_INSN1(cond_seq, line, splatarray, Qfalse);
ADD_INSN1(cond_seq, line, checkmatch, INT2FIX(VM_CHECKMATCH_TYPE_CASE | VM_CHECKMATCH_ARRAY));
ADD_INSNL(cond_seq, line, branchif, l1);
break;
}
return COMPILE_OK;
}
static int
compile_massign_lhs(rb_iseq_t *iseq, LINK_ANCHOR *const ret, const NODE *const node)
{
switch (nd_type(node)) {
case NODE_ATTRASGN: {
INSN *iobj;
struct rb_call_info *ci;
VALUE dupidx;
int line = nd_line(node);
CHECK(COMPILE_POPPED(ret, "masgn lhs (NODE_ATTRASGN)", node));
iobj = (INSN *)get_prev_insn((INSN *)LAST_ELEMENT(ret)); /* send insn */
ci = (struct rb_call_info *)OPERAND_AT(iobj, 0);
ci->orig_argc += 1;
dupidx = INT2FIX(ci->orig_argc);
INSERT_BEFORE_INSN1(iobj, line, topn, dupidx);
if (ci->flag & VM_CALL_ARGS_SPLAT) {
--ci->orig_argc;
INSERT_BEFORE_INSN1(iobj, line, newarray, INT2FIX(1));
INSERT_BEFORE_INSN(iobj, line, concatarray);
}
ADD_INSN(ret, line, pop); /* result */
break;
}
case NODE_MASGN: {
DECL_ANCHOR(anchor);
INIT_ANCHOR(anchor);
CHECK(COMPILE_POPPED(anchor, "nest masgn lhs", node));
ELEM_REMOVE(FIRST_ELEMENT(anchor));
ADD_SEQ(ret, anchor);
break;
}
default: {
DECL_ANCHOR(anchor);
INIT_ANCHOR(anchor);
CHECK(COMPILE_POPPED(anchor, "masgn lhs", node));
ELEM_REMOVE(FIRST_ELEMENT(anchor));
ADD_SEQ(ret, anchor);
}
}
return COMPILE_OK;
}
static int
compile_massign_opt_lhs(rb_iseq_t *iseq, LINK_ANCHOR *const ret, const NODE *lhsn)
{
if (lhsn) {
CHECK(compile_massign_opt_lhs(iseq, ret, lhsn->nd_next));
CHECK(compile_massign_lhs(iseq, ret, lhsn->nd_head));
}
return COMPILE_OK;
}
static int
compile_massign_opt(rb_iseq_t *iseq, LINK_ANCHOR *const ret,
const NODE *rhsn, const NODE *orig_lhsn)
{
VALUE mem[64];
const int memsize = numberof(mem);
int memindex = 0;
int llen = 0, rlen = 0;
int i;
const NODE *lhsn = orig_lhsn;
#define MEMORY(v) { \
int i; \
if (memindex == memsize) return 0; \
for (i=0; i<memindex; i++) { \
if (mem[i] == (v)) return 0; \
} \
mem[memindex++] = (v); \
}
if (rhsn == 0 || nd_type(rhsn) != NODE_LIST) {
return 0;
}
while (lhsn) {
const NODE *ln = lhsn->nd_head;
switch (nd_type(ln)) {
case NODE_LASGN:
MEMORY(ln->nd_vid);
break;
case NODE_DASGN:
case NODE_DASGN_CURR:
case NODE_IASGN:
case NODE_CVASGN:
MEMORY(ln->nd_vid);
break;
default:
return 0;
}
lhsn = lhsn->nd_next;
llen++;
}
while (rhsn) {
if (llen <= rlen) {
2019-07-15 08:43:39 +03:00
NO_CHECK(COMPILE_POPPED(ret, "masgn val (popped)", rhsn->nd_head));
}
else {
2019-07-15 08:43:39 +03:00
NO_CHECK(COMPILE(ret, "masgn val", rhsn->nd_head));
}
rhsn = rhsn->nd_next;
rlen++;
}
if (llen > rlen) {
for (i=0; i<llen-rlen; i++) {
ADD_INSN(ret, nd_line(orig_lhsn), putnil);
}
}
compile_massign_opt_lhs(iseq, ret, orig_lhsn);
return 1;
}
static void
adjust_stack(rb_iseq_t *iseq, LINK_ANCHOR *const ret, int line, int rlen, int llen)
{
if (rlen < llen) {
do {ADD_INSN(ret, line, putnil);} while (++rlen < llen);
}
else if (rlen > llen) {
do {ADD_INSN(ret, line, pop);} while (--rlen > llen);
}
}
static int
compile_massign(rb_iseq_t *iseq, LINK_ANCHOR *const ret, const NODE *const node, int popped)
{
const NODE *rhsn = node->nd_value;
const NODE *splatn = node->nd_args;
const NODE *lhsn = node->nd_head;
int lhs_splat = (splatn && NODE_NAMED_REST_P(splatn)) ? 1 : 0;
if (!popped || splatn || !compile_massign_opt(iseq, ret, rhsn, lhsn)) {
int llen = 0;
int expand = 1;
DECL_ANCHOR(lhsseq);
INIT_ANCHOR(lhsseq);
while (lhsn) {
CHECK(compile_massign_lhs(iseq, lhsseq, lhsn->nd_head));
llen += 1;
lhsn = lhsn->nd_next;
}
2019-07-15 08:43:39 +03:00
NO_CHECK(COMPILE(ret, "normal masgn rhs", rhsn));
if (!popped) {
ADD_INSN(ret, nd_line(node), dup);
}
else if (!lhs_splat) {
INSN *last = (INSN*)ret->last;
if (IS_INSN(&last->link) &&
IS_INSN_ID(last, newarray) &&
last->operand_size == 1) {
int rlen = FIX2INT(OPERAND_AT(last, 0));
/* special case: assign to aset or attrset */
if (llen == 2) {
POP_ELEMENT(ret);
adjust_stack(iseq, ret, nd_line(node), rlen, llen);
ADD_INSN(ret, nd_line(node), swap);
expand = 0;
}
else if (llen > 2 && llen != rlen) {
POP_ELEMENT(ret);
adjust_stack(iseq, ret, nd_line(node), rlen, llen);
ADD_INSN1(ret, nd_line(node), reverse, INT2FIX(llen));
expand = 0;
}
else if (llen > 2) {
last->insn_id = BIN(reverse);
expand = 0;
}
}
}
if (expand) {
ADD_INSN2(ret, nd_line(node), expandarray,
INT2FIX(llen), INT2FIX(lhs_splat));
}
ADD_SEQ(ret, lhsseq);
if (lhs_splat) {
if (nd_type(splatn) == NODE_POSTARG) {
/*a, b, *r, p1, p2 */
const NODE *postn = splatn->nd_2nd;
const NODE *restn = splatn->nd_1st;
int num = (int)postn->nd_alen;
int flag = 0x02 | (NODE_NAMED_REST_P(restn) ? 0x01 : 0x00);
ADD_INSN2(ret, nd_line(splatn), expandarray,
INT2FIX(num), INT2FIX(flag));
if (NODE_NAMED_REST_P(restn)) {
CHECK(compile_massign_lhs(iseq, ret, restn));
}
while (postn) {
CHECK(compile_massign_lhs(iseq, ret, postn->nd_head));
postn = postn->nd_next;
}
}
else {
/* a, b, *r */
CHECK(compile_massign_lhs(iseq, ret, splatn));
}
}
}
return COMPILE_OK;
}
static int
compile_const_prefix(rb_iseq_t *iseq, const NODE *const node,
LINK_ANCHOR *const pref, LINK_ANCHOR *const body)
{
switch (nd_type(node)) {
case NODE_CONST:
debugi("compile_const_prefix - colon", node->nd_vid);
ADD_INSN1(body, nd_line(node), putobject, Qtrue);
ADD_INSN1(body, nd_line(node), getconstant, ID2SYM(node->nd_vid));
break;
case NODE_COLON3:
debugi("compile_const_prefix - colon3", node->nd_mid);
ADD_INSN(body, nd_line(node), pop);
ADD_INSN1(body, nd_line(node), putobject, rb_cObject);
ADD_INSN1(body, nd_line(node), putobject, Qtrue);
ADD_INSN1(body, nd_line(node), getconstant, ID2SYM(node->nd_mid));
break;
case NODE_COLON2:
CHECK(compile_const_prefix(iseq, node->nd_head, pref, body));
debugi("compile_const_prefix - colon2", node->nd_mid);
ADD_INSN1(body, nd_line(node), putobject, Qfalse);
ADD_INSN1(body, nd_line(node), getconstant, ID2SYM(node->nd_mid));
break;
default:
CHECK(COMPILE(pref, "const colon2 prefix", node));
break;
}
return COMPILE_OK;
}
static int
compile_cpath(LINK_ANCHOR *const ret, rb_iseq_t *iseq, const NODE *cpath)
{
if (nd_type(cpath) == NODE_COLON3) {
/* toplevel class ::Foo */
ADD_INSN1(ret, nd_line(cpath), putobject, rb_cObject);
return VM_DEFINECLASS_FLAG_SCOPED;
}
else if (cpath->nd_head) {
/* Bar::Foo */
2019-07-15 08:43:39 +03:00
NO_CHECK(COMPILE(ret, "nd_else->nd_head", cpath->nd_head));
return VM_DEFINECLASS_FLAG_SCOPED;
}
else {
/* class at cbase Foo */
ADD_INSN1(ret, nd_line(cpath), putspecialobject,
INT2FIX(VM_SPECIAL_OBJECT_CONST_BASE));
return 0;
}
}
static inline int
private_recv_p(const NODE *node)
{
if (nd_type(node->nd_recv) == NODE_SELF) {
NODE *self = node->nd_recv;
return self->nd_state != 0;
}
return 0;
}
static void
defined_expr(rb_iseq_t *iseq, LINK_ANCHOR *const ret,
const NODE *const node, LABEL **lfinish, VALUE needstr);
static void
defined_expr0(rb_iseq_t *iseq, LINK_ANCHOR *const ret,
const NODE *const node, LABEL **lfinish, VALUE needstr)
{
enum defined_type expr_type = DEFINED_NOT_DEFINED;
enum node_type type;
const int line = nd_line(node);
switch (type = nd_type(node)) {
/* easy literals */
case NODE_NIL:
expr_type = DEFINED_NIL;
break;
case NODE_SELF:
expr_type = DEFINED_SELF;
break;
case NODE_TRUE:
expr_type = DEFINED_TRUE;
break;
case NODE_FALSE:
expr_type = DEFINED_FALSE;
break;
case NODE_LIST:{
const NODE *vals = node;
do {
defined_expr0(iseq, ret, vals->nd_head, lfinish, Qfalse);
if (!lfinish[1]) {
lfinish[1] = NEW_LABEL(line);
}
ADD_INSNL(ret, line, branchunless, lfinish[1]);
} while ((vals = vals->nd_next) != NULL);
}
2019-07-15 01:56:53 +03:00
/* fall through */
case NODE_STR:
case NODE_LIT:
case NODE_ZLIST:
case NODE_AND:
case NODE_OR:
default:
expr_type = DEFINED_EXPR;
break;
/* variables */
case NODE_LVAR:
case NODE_DVAR:
expr_type = DEFINED_LVAR;
break;
case NODE_IVAR:
ADD_INSN(ret, line, putnil);
ADD_INSN3(ret, line, defined, INT2FIX(DEFINED_IVAR),
ID2SYM(node->nd_vid), needstr);
2019-07-14 10:18:48 +03:00
return;
case NODE_GVAR:
ADD_INSN(ret, line, putnil);
ADD_INSN3(ret, line, defined, INT2FIX(DEFINED_GVAR),
ID2SYM(node->nd_entry->id), needstr);
2019-07-14 10:18:48 +03:00
return;
case NODE_CVAR:
ADD_INSN(ret, line, putnil);
ADD_INSN3(ret, line, defined, INT2FIX(DEFINED_CVAR),
ID2SYM(node->nd_vid), needstr);
2019-07-14 10:18:48 +03:00
return;
case NODE_CONST:
ADD_INSN(ret, line, putnil);
ADD_INSN3(ret, line, defined, INT2FIX(DEFINED_CONST),
ID2SYM(node->nd_vid), needstr);
2019-07-14 10:18:48 +03:00
return;
case NODE_COLON2:
if (!lfinish[1]) {
lfinish[1] = NEW_LABEL(line);
}
defined_expr0(iseq, ret, node->nd_head, lfinish, Qfalse);
ADD_INSNL(ret, line, branchunless, lfinish[1]);
2019-07-15 08:43:39 +03:00
NO_CHECK(COMPILE(ret, "defined/colon2#nd_head", node->nd_head));
ADD_INSN3(ret, line, defined,
(rb_is_const_id(node->nd_mid) ?
INT2FIX(DEFINED_CONST) : INT2FIX(DEFINED_METHOD)),
ID2SYM(node->nd_mid), needstr);
2019-07-14 10:18:48 +03:00
return;
case NODE_COLON3:
ADD_INSN1(ret, line, putobject, rb_cObject);
ADD_INSN3(ret, line, defined,
INT2FIX(DEFINED_CONST), ID2SYM(node->nd_mid), needstr);
2019-07-14 10:18:48 +03:00
return;
/* method dispatch */
case NODE_CALL:
case NODE_OPCALL:
case NODE_VCALL:
case NODE_FCALL:
case NODE_METHREF:
case NODE_ATTRASGN:{
const int explicit_receiver =
(type == NODE_CALL || type == NODE_OPCALL ||
type == NODE_METHREF ||
(type == NODE_ATTRASGN && !private_recv_p(node)));
if (!lfinish[1] && (node->nd_args || explicit_receiver)) {
lfinish[1] = NEW_LABEL(line);
}
if (node->nd_args) {
defined_expr0(iseq, ret, node->nd_args, lfinish, Qfalse);
ADD_INSNL(ret, line, branchunless, lfinish[1]);
}
if (explicit_receiver) {
defined_expr0(iseq, ret, node->nd_recv, lfinish, Qfalse);
ADD_INSNL(ret, line, branchunless, lfinish[1]);
2019-07-15 08:43:39 +03:00
NO_CHECK(COMPILE(ret, "defined/recv", node->nd_recv));
ADD_INSN3(ret, line, defined, INT2FIX(DEFINED_METHOD),
ID2SYM(node->nd_mid), needstr);
}
else {
ADD_INSN(ret, line, putself);
ADD_INSN3(ret, line, defined, INT2FIX(DEFINED_FUNC),
ID2SYM(node->nd_mid), needstr);
}
2019-07-14 10:18:48 +03:00
return;
}
case NODE_YIELD:
ADD_INSN(ret, line, putnil);
ADD_INSN3(ret, line, defined, INT2FIX(DEFINED_YIELD), 0,
needstr);
2019-07-14 10:18:48 +03:00
return;
case NODE_BACK_REF:
case NODE_NTH_REF:
ADD_INSN(ret, line, putnil);
ADD_INSN3(ret, line, defined, INT2FIX(DEFINED_REF),
INT2FIX((node->nd_nth << 1) | (type == NODE_BACK_REF)),
needstr);
2019-07-14 10:18:48 +03:00
return;
case NODE_SUPER:
case NODE_ZSUPER:
ADD_INSN(ret, line, putnil);
ADD_INSN3(ret, line, defined, INT2FIX(DEFINED_ZSUPER), 0,
needstr);
2019-07-14 10:18:48 +03:00
return;
case NODE_OP_ASGN1:
case NODE_OP_ASGN2:
case NODE_OP_ASGN_OR:
case NODE_OP_ASGN_AND:
case NODE_MASGN:
case NODE_LASGN:
case NODE_DASGN:
case NODE_DASGN_CURR:
case NODE_GASGN:
case NODE_IASGN:
case NODE_CDECL:
case NODE_CVASGN:
expr_type = DEFINED_ASGN;
break;
}
assert(expr_type != DEFINED_NOT_DEFINED);
if (needstr != Qfalse) {
VALUE str = rb_iseq_defined_string(expr_type);
ADD_INSN1(ret, line, putobject, str);
}
else {
ADD_INSN1(ret, line, putobject, Qtrue);
}
}
static void
build_defined_rescue_iseq(rb_iseq_t *iseq, LINK_ANCHOR *const ret, const void *unused)
{
ADD_INSN(ret, 0, putnil);
iseq_set_exception_local_table(iseq);
}
static void
defined_expr(rb_iseq_t *iseq, LINK_ANCHOR *const ret,
const NODE *const node, LABEL **lfinish, VALUE needstr)
{
LINK_ELEMENT *lcur = ret->last;
defined_expr0(iseq, ret, node, lfinish, needstr);
if (lfinish[1]) {
int line = nd_line(node);
LABEL *lstart = NEW_LABEL(line);
LABEL *lend = NEW_LABEL(line);
const rb_iseq_t *rescue;
struct rb_iseq_new_with_callback_callback_func *ifunc =
rb_iseq_new_with_callback_new_callback(build_defined_rescue_iseq, NULL);
rescue = new_child_iseq_with_callback(iseq, ifunc,
rb_str_concat(rb_str_new2("defined guard in "),
iseq->body->location.label),
iseq, ISEQ_TYPE_RESCUE, 0);
lstart->rescued = LABEL_RESCUE_BEG;
lend->rescued = LABEL_RESCUE_END;
APPEND_LABEL(ret, lcur, lstart);
ADD_LABEL(ret, lend);
ADD_CATCH_ENTRY(CATCH_TYPE_RESCUE, lstart, lend, rescue, lfinish[1]);
}
}
static int
compile_defined_expr(rb_iseq_t *iseq, LINK_ANCHOR *const ret, const NODE *const node, VALUE needstr)
{
const int line = nd_line(node);
if (!node->nd_head) {
VALUE str = rb_iseq_defined_string(DEFINED_NIL);
ADD_INSN1(ret, line, putobject, str);
}
else {
LABEL *lfinish[2];
LINK_ELEMENT *last = ret->last;
lfinish[0] = NEW_LABEL(line);
lfinish[1] = 0;
defined_expr(iseq, ret, node->nd_head, lfinish, needstr);
if (lfinish[1]) {
ELEM_INSERT_NEXT(last, &new_insn_body(iseq, line, BIN(putnil), 0)->link);
ADD_INSN(ret, line, swap);
ADD_INSN(ret, line, pop);
ADD_LABEL(ret, lfinish[1]);
}
ADD_LABEL(ret, lfinish[0]);
}
return COMPILE_OK;
}
static VALUE
2015-07-22 01:52:59 +03:00
make_name_for_block(const rb_iseq_t *orig_iseq)
{
int level = 1;
2015-07-22 01:52:59 +03:00
const rb_iseq_t *iseq = orig_iseq;
2015-07-22 01:52:59 +03:00
if (orig_iseq->body->parent_iseq != 0) {
while (orig_iseq->body->local_iseq != iseq) {
if (iseq->body->type == ISEQ_TYPE_BLOCK) {
level++;
}
2015-07-22 01:52:59 +03:00
iseq = iseq->body->parent_iseq;
}
}
if (level == 1) {
2015-07-22 01:52:59 +03:00
return rb_sprintf("block in %"PRIsVALUE, iseq->body->location.label);
}
else {
2015-07-22 01:52:59 +03:00
return rb_sprintf("block (%d levels) in %"PRIsVALUE, level, iseq->body->location.label);
}
}
static void
push_ensure_entry(rb_iseq_t *iseq,
struct iseq_compile_data_ensure_node_stack *enl,
struct ensure_range *er, const NODE *const node)
{
enl->ensure_node = node;
enl->prev = ISEQ_COMPILE_DATA(iseq)->ensure_node_stack; /* prev */
enl->erange = er;
ISEQ_COMPILE_DATA(iseq)->ensure_node_stack = enl;
}
static void
add_ensure_range(rb_iseq_t *iseq, struct ensure_range *erange,
LABEL *lstart, LABEL *lend)
{
struct ensure_range *ne =
compile_data_alloc(iseq, sizeof(struct ensure_range));
while (erange->next != 0) {
erange = erange->next;
}
ne->next = 0;
ne->begin = lend;
ne->end = erange->end;
erange->end = lstart;
erange->next = ne;
}
static void
add_ensure_iseq(LINK_ANCHOR *const ret, rb_iseq_t *iseq, int is_return)
{
struct iseq_compile_data_ensure_node_stack *enlp =
ISEQ_COMPILE_DATA(iseq)->ensure_node_stack;
struct iseq_compile_data_ensure_node_stack *prev_enlp = enlp;
DECL_ANCHOR(ensure);
INIT_ANCHOR(ensure);
while (enlp) {
if (enlp->erange != NULL) {
DECL_ANCHOR(ensure_part);
LABEL *lstart = NEW_LABEL(0);
LABEL *lend = NEW_LABEL(0);
INIT_ANCHOR(ensure_part);
add_ensure_range(iseq, enlp->erange, lstart, lend);
ISEQ_COMPILE_DATA(iseq)->ensure_node_stack = enlp->prev;
ADD_LABEL(ensure_part, lstart);
2019-07-15 08:43:39 +03:00
NO_CHECK(COMPILE_POPPED(ensure_part, "ensure part", enlp->ensure_node));
ADD_LABEL(ensure_part, lend);
ADD_SEQ(ensure, ensure_part);
}
else {
if (!is_return) {
break;
}
}
enlp = enlp->prev;
}
ISEQ_COMPILE_DATA(iseq)->ensure_node_stack = prev_enlp;
ADD_SEQ(ret, ensure);
}
static int
check_keyword(const NODE *node)
{
/* This check is essentially a code clone of compile_keyword_arg. */
if (nd_type(node) == NODE_LIST) {
while (node->nd_next) {
node = node->nd_next;
}
node = node->nd_head;
}
return keyword_node_p(node);
}
static VALUE
setup_args_core(rb_iseq_t *iseq, LINK_ANCHOR *const args, const NODE *argn,
int dup_rest, unsigned int *flag, struct rb_call_info_kw_arg **keywords)
{
if (argn) {
switch (nd_type(argn)) {
case NODE_SPLAT: {
NO_CHECK(COMPILE(args, "args (splat)", argn->nd_head));
ADD_INSN1(args, nd_line(argn), splatarray, dup_rest ? Qtrue : Qfalse);
if (flag) *flag |= VM_CALL_ARGS_SPLAT;
return INT2FIX(1);
}
case NODE_ARGSCAT:
case NODE_ARGSPUSH: {
int next_is_list = (nd_type(argn->nd_head) == NODE_LIST);
VALUE argc = setup_args_core(iseq, args, argn->nd_head, 1, NULL, NULL);
NO_CHECK(COMPILE(args, "args (cat: splat)", argn->nd_body));
if (flag) {
*flag |= VM_CALL_ARGS_SPLAT;
/* This is a dirty hack. It traverses the AST twice.
* In a long term, it should be fixed by a redesign of keyword arguments */
if (check_keyword(argn->nd_body))
*flag |= VM_CALL_KW_SPLAT;
}
if (nd_type(argn) == NODE_ARGSCAT) {
if (next_is_list) {
ADD_INSN1(args, nd_line(argn), splatarray, Qtrue);
return INT2FIX(FIX2INT(argc) + 1);
}
else {
ADD_INSN1(args, nd_line(argn), splatarray, Qfalse);
ADD_INSN(args, nd_line(argn), concatarray);
return argc;
}
}
else {
ADD_INSN1(args, nd_line(argn), newarray, INT2FIX(1));
ADD_INSN(args, nd_line(argn), concatarray);
return argc;
}
}
case NODE_LIST: {
int len = compile_args(iseq, args, argn, keywords, flag);
return INT2FIX(len);
}
default: {
UNKNOWN_NODE("setup_arg", argn, Qnil);
}
}
}
return INT2FIX(0);
}
static VALUE
setup_args(rb_iseq_t *iseq, LINK_ANCHOR *const args, const NODE *argn,
unsigned int *flag, struct rb_call_info_kw_arg **keywords)
{
VALUE ret;
if (argn && nd_type(argn) == NODE_BLOCK_PASS) {
DECL_ANCHOR(arg_block);
INIT_ANCHOR(arg_block);
NO_CHECK(COMPILE(arg_block, "block", argn->nd_body));
*flag |= VM_CALL_ARGS_BLOCKARG;
ret = setup_args_core(iseq, args, argn->nd_head, 0, flag, keywords);
if (LIST_INSN_SIZE_ONE(arg_block)) {
LINK_ELEMENT *elem = FIRST_ELEMENT(arg_block);
if (elem->type == ISEQ_ELEMENT_INSN) {
INSN *iobj = (INSN *)elem;
if (iobj->insn_id == BIN(getblockparam)) {
iobj->insn_id = BIN(getblockparamproxy);
}
}
}
ADD_SEQ(args, arg_block);
}
else {
ret = setup_args_core(iseq, args, argn, 0, flag, keywords);
}
return ret;
}
static void
build_postexe_iseq(rb_iseq_t *iseq, LINK_ANCHOR *ret, const void *ptr)
{
const NODE *body = ptr;
int line = nd_line(body);
VALUE argc = INT2FIX(0);
2015-07-22 01:52:59 +03:00
const rb_iseq_t *block = NEW_CHILD_ISEQ(body, make_name_for_block(iseq->body->parent_iseq), ISEQ_TYPE_BLOCK, line);
ADD_INSN1(ret, line, putspecialobject, INT2FIX(VM_SPECIAL_OBJECT_VMCORE));
ADD_CALL_WITH_BLOCK(ret, line, id_core_set_postexe, argc, block);
RB_OBJ_WRITTEN(iseq, Qundef, (VALUE)block);
iseq_set_local_table(iseq, 0);
}
static void
compile_named_capture_assign(rb_iseq_t *iseq, LINK_ANCHOR *const ret, const NODE *const node)
{
const NODE *vars;
LINK_ELEMENT *last;
int line = nd_line(node);
LABEL *fail_label = NEW_LABEL(line), *end_label = NEW_LABEL(line);
#if !(defined(NAMED_CAPTURE_BY_SVAR) && NAMED_CAPTURE_BY_SVAR-0)
ADD_INSN1(ret, line, getglobal, ((VALUE)rb_global_entry(idBACKREF) | 1));
#else
ADD_INSN2(ret, line, getspecial, INT2FIX(1) /* '~' */, INT2FIX(0));
#endif
ADD_INSN(ret, line, dup);
ADD_INSNL(ret, line, branchunless, fail_label);
for (vars = node; vars; vars = vars->nd_next) {
INSN *cap;
if (vars->nd_next) {
ADD_INSN(ret, line, dup);
}
last = ret->last;
2019-07-15 08:43:39 +03:00
NO_CHECK(COMPILE_POPPED(ret, "capture", vars->nd_head));
last = last->next; /* putobject :var */
cap = new_insn_send(iseq, line, idAREF, INT2FIX(1),
NULL, INT2FIX(0), NULL);
ELEM_INSERT_PREV(last->next, (LINK_ELEMENT *)cap);
#if !defined(NAMED_CAPTURE_SINGLE_OPT) || NAMED_CAPTURE_SINGLE_OPT-0
if (!vars->nd_next && vars == node) {
/* only one name */
DECL_ANCHOR(nom);
INIT_ANCHOR(nom);
ADD_INSNL(nom, line, jump, end_label);
ADD_LABEL(nom, fail_label);
# if 0 /* $~ must be MatchData or nil */
ADD_INSN(nom, line, pop);
ADD_INSN(nom, line, putnil);
# endif
ADD_LABEL(nom, end_label);
(nom->last->next = cap->link.next)->prev = nom->last;
(cap->link.next = nom->anchor.next)->prev = &cap->link;
return;
}
#endif
}
ADD_INSNL(ret, line, jump, end_label);
ADD_LABEL(ret, fail_label);
ADD_INSN(ret, line, pop);
for (vars = node; vars; vars = vars->nd_next) {
last = ret->last;
2019-07-16 01:19:52 +03:00
NO_CHECK(COMPILE_POPPED(ret, "capture", vars->nd_head));
last = last->next; /* putobject :var */
((INSN*)last)->insn_id = BIN(putnil);
((INSN*)last)->operand_size = 0;
}
ADD_LABEL(ret, end_label);
}
static int
number_literal_p(const NODE *n)
{
return (n && nd_type(n) == NODE_LIT && RB_INTEGER_TYPE_P(n->nd_lit));
}
static int
compile_if(rb_iseq_t *iseq, LINK_ANCHOR *const ret, const NODE *const node, int popped, const enum node_type type)
{
struct rb_iseq_constant_body *const body = iseq->body;
const NODE *const node_body = type == NODE_IF ? node->nd_body : node->nd_else;
const NODE *const node_else = type == NODE_IF ? node->nd_else : node->nd_body;
const int line = nd_line(node);
const int lineno = nd_first_lineno(node);
const int column = nd_first_column(node);
const int last_lineno = nd_last_lineno(node);
const int last_column = nd_last_column(node);
DECL_ANCHOR(cond_seq);
DECL_ANCHOR(then_seq);
DECL_ANCHOR(else_seq);
LABEL *then_label, *else_label, *end_label;
VALUE branches = Qfalse;
int ci_size, ci_kw_size;
VALUE catch_table = ISEQ_COMPILE_DATA(iseq)->catch_table_ary;
long catch_table_size = NIL_P(catch_table) ? 0 : RARRAY_LEN(catch_table);
INIT_ANCHOR(cond_seq);
INIT_ANCHOR(then_seq);
INIT_ANCHOR(else_seq);
then_label = NEW_LABEL(line);
else_label = NEW_LABEL(line);
end_label = 0;
compile_branch_condition(iseq, cond_seq, node->nd_cond,
then_label, else_label);
ci_size = body->ci_size;
ci_kw_size = body->ci_kw_size;
CHECK(COMPILE_(then_seq, "then", node_body, popped));
catch_table = ISEQ_COMPILE_DATA(iseq)->catch_table_ary;
if (!then_label->refcnt) {
body->ci_size = ci_size;
body->ci_kw_size = ci_kw_size;
if (!NIL_P(catch_table)) rb_ary_set_len(catch_table, catch_table_size);
}
else {
if (!NIL_P(catch_table)) catch_table_size = RARRAY_LEN(catch_table);
}
ci_size = body->ci_size;
ci_kw_size = body->ci_kw_size;
CHECK(COMPILE_(else_seq, "else", node_else, popped));
catch_table = ISEQ_COMPILE_DATA(iseq)->catch_table_ary;
if (!else_label->refcnt) {
body->ci_size = ci_size;
body->ci_kw_size = ci_kw_size;
if (!NIL_P(catch_table)) rb_ary_set_len(catch_table, catch_table_size);
}
else {
if (!NIL_P(catch_table)) catch_table_size = RARRAY_LEN(catch_table);
}
ADD_SEQ(ret, cond_seq);
if (then_label->refcnt && else_label->refcnt) {
DECL_BRANCH_BASE(branches, lineno, column, last_lineno, last_column, type == NODE_IF ? "if" : "unless");
}
if (then_label->refcnt) {
ADD_LABEL(ret, then_label);
if (else_label->refcnt) {
ADD_TRACE_BRANCH_COVERAGE(
ret,
node_body ? nd_first_lineno(node_body) : lineno,
node_body ? nd_first_column(node_body) : column,
node_body ? nd_last_lineno(node_body) : last_lineno,
node_body ? nd_last_column(node_body) : last_column,
type == NODE_IF ? "then" : "else",
branches);
end_label = NEW_LABEL(line);
ADD_INSNL(then_seq, line, jump, end_label);
}
ADD_SEQ(ret, then_seq);
}
if (else_label->refcnt) {
ADD_LABEL(ret, else_label);
if (then_label->refcnt) {
ADD_TRACE_BRANCH_COVERAGE(
ret,
node_else ? nd_first_lineno(node_else) : lineno,
node_else ? nd_first_column(node_else) : column,
node_else ? nd_last_lineno(node_else) : last_lineno,
node_else ? nd_last_column(node_else) : last_column,
type == NODE_IF ? "else" : "then",
branches);
}
ADD_SEQ(ret, else_seq);
}
if (end_label) {
ADD_LABEL(ret, end_label);
}
return COMPILE_OK;
}
static int
compile_case(rb_iseq_t *iseq, LINK_ANCHOR *const ret, const NODE *const orig_node, int popped)
{
const NODE *vals;
const NODE *node = orig_node;
LABEL *endlabel, *elselabel;
DECL_ANCHOR(head);
DECL_ANCHOR(body_seq);
DECL_ANCHOR(cond_seq);
int only_special_literals = 1;
VALUE literals = rb_hash_new();
int line, lineno, column, last_lineno, last_column;
enum node_type type;
VALUE branches = Qfalse;
INIT_ANCHOR(head);
INIT_ANCHOR(body_seq);
INIT_ANCHOR(cond_seq);
RHASH_TBL_RAW(literals)->type = &cdhash_type;
CHECK(COMPILE(head, "case base", node->nd_head));
DECL_BRANCH_BASE(branches, nd_first_lineno(node), nd_first_column(node), nd_last_lineno(node), nd_last_column(node), "case");
node = node->nd_body;
EXPECT_NODE("NODE_CASE", node, NODE_WHEN, COMPILE_NG);
type = nd_type(node);
line = nd_line(node);
lineno = nd_first_lineno(node);
column = nd_first_column(node);
last_lineno = nd_last_lineno(node);
last_column = nd_last_column(node);
endlabel = NEW_LABEL(line);
elselabel = NEW_LABEL(line);
ADD_SEQ(ret, head); /* case VAL */
while (type == NODE_WHEN) {
LABEL *l1;
l1 = NEW_LABEL(line);
ADD_LABEL(body_seq, l1);
ADD_INSN(body_seq, line, pop);
ADD_TRACE_BRANCH_COVERAGE(
body_seq,
node->nd_body ? nd_first_lineno(node->nd_body) : lineno,
node->nd_body ? nd_first_column(node->nd_body) : column,
node->nd_body ? nd_last_lineno(node->nd_body) : last_lineno,
node->nd_body ? nd_last_column(node->nd_body) : last_column,
"when",
branches);
CHECK(COMPILE_(body_seq, "when body", node->nd_body, popped));
ADD_INSNL(body_seq, line, jump, endlabel);
vals = node->nd_head;
if (vals) {
switch (nd_type(vals)) {
case NODE_LIST:
only_special_literals = when_vals(iseq, cond_seq, vals, l1, only_special_literals, literals);
if (only_special_literals < 0) return COMPILE_NG;
break;
case NODE_SPLAT:
case NODE_ARGSCAT:
case NODE_ARGSPUSH:
only_special_literals = 0;
CHECK(when_splat_vals(iseq, cond_seq, vals, l1, only_special_literals, literals));
break;
default:
UNKNOWN_NODE("NODE_CASE", vals, COMPILE_NG);
}
}
else {
EXPECT_NODE_NONULL("NODE_CASE", node, NODE_LIST, COMPILE_NG);
}
node = node->nd_next;
if (!node) {
break;
}
type = nd_type(node);
line = nd_line(node);
lineno = nd_first_lineno(node);
column = nd_first_column(node);
last_lineno = nd_last_lineno(node);
last_column = nd_last_column(node);
}
/* else */
if (node) {
ADD_LABEL(cond_seq, elselabel);
ADD_INSN(cond_seq, line, pop);
ADD_TRACE_BRANCH_COVERAGE(cond_seq, nd_first_lineno(node), nd_first_column(node), nd_last_lineno(node), nd_last_column(node), "else", branches);
CHECK(COMPILE_(cond_seq, "else", node, popped));
ADD_INSNL(cond_seq, line, jump, endlabel);
}
else {
debugs("== else (implicit)\n");
ADD_LABEL(cond_seq, elselabel);
ADD_INSN(cond_seq, nd_line(orig_node), pop);
ADD_TRACE_BRANCH_COVERAGE(cond_seq, nd_first_lineno(orig_node), nd_first_column(orig_node), nd_last_lineno(orig_node), nd_last_column(orig_node), "else", branches);
if (!popped) {
ADD_INSN(cond_seq, nd_line(orig_node), putnil);
}
ADD_INSNL(cond_seq, nd_line(orig_node), jump, endlabel);
}
if (only_special_literals && ISEQ_COMPILE_DATA(iseq)->option->specialized_instruction) {
ADD_INSN(ret, nd_line(orig_node), dup);
ADD_INSN2(ret, nd_line(orig_node), opt_case_dispatch, literals, elselabel);
RB_OBJ_WRITTEN(iseq, Qundef, literals);
LABEL_REF(elselabel);
}
ADD_SEQ(ret, cond_seq);
ADD_SEQ(ret, body_seq);
ADD_LABEL(ret, endlabel);
return COMPILE_OK;
}
static int
compile_case2(rb_iseq_t *iseq, LINK_ANCHOR *const ret, const NODE *const orig_node, int popped)
{
const NODE *vals;
const NODE *val;
const NODE *node = orig_node->nd_body;
LABEL *endlabel;
DECL_ANCHOR(body_seq);
VALUE branches = Qfalse;
DECL_BRANCH_BASE(branches, nd_first_lineno(orig_node), nd_first_column(orig_node), nd_last_lineno(orig_node), nd_last_column(orig_node), "case");
INIT_ANCHOR(body_seq);
endlabel = NEW_LABEL(nd_line(node));
while (node && nd_type(node) == NODE_WHEN) {
const int line = nd_line(node);
const int lineno = nd_first_lineno(node);
const int column = nd_first_column(node);
const int last_lineno = nd_last_lineno(node);
const int last_column = nd_last_column(node);
LABEL *l1 = NEW_LABEL(line);
ADD_LABEL(body_seq, l1);
ADD_TRACE_BRANCH_COVERAGE(
body_seq,
node->nd_body ? nd_first_lineno(node->nd_body) : lineno,
node->nd_body ? nd_first_column(node->nd_body) : column,
node->nd_body ? nd_last_lineno(node->nd_body) : last_lineno,
node->nd_body ? nd_last_column(node->nd_body) : last_column,
"when",
branches);
CHECK(COMPILE_(body_seq, "when", node->nd_body, popped));
ADD_INSNL(body_seq, line, jump, endlabel);
vals = node->nd_head;
if (!vals) {
2019-09-19 17:45:09 +03:00
EXPECT_NODE_NONULL("NODE_WHEN", node, NODE_LIST, COMPILE_NG);
}
switch (nd_type(vals)) {
case NODE_LIST:
while (vals) {
LABEL *lnext;
val = vals->nd_head;
lnext = NEW_LABEL(nd_line(val));
debug_compile("== when2\n", (void)0);
CHECK(compile_branch_condition(iseq, ret, val, l1, lnext));
ADD_LABEL(ret, lnext);
vals = vals->nd_next;
}
break;
case NODE_SPLAT:
case NODE_ARGSCAT:
case NODE_ARGSPUSH:
ADD_INSN(ret, nd_line(vals), putnil);
CHECK(COMPILE(ret, "when2/cond splat", vals));
ADD_INSN1(ret, nd_line(vals), checkmatch, INT2FIX(VM_CHECKMATCH_TYPE_WHEN | VM_CHECKMATCH_ARRAY));
ADD_INSNL(ret, nd_line(vals), branchif, l1);
break;
default:
UNKNOWN_NODE("NODE_WHEN", vals, COMPILE_NG);
}
node = node->nd_next;
}
/* else */
ADD_TRACE_BRANCH_COVERAGE(
ret,
node ? nd_first_lineno(node) : nd_first_lineno(orig_node),
node ? nd_first_column(node) : nd_first_column(orig_node),
node ? nd_last_lineno(node) : nd_last_lineno(orig_node),
node ? nd_last_column(node) : nd_last_column(orig_node),
"else",
branches);
CHECK(COMPILE_(ret, "else", node, popped));
ADD_INSNL(ret, nd_line(orig_node), jump, endlabel);
ADD_SEQ(ret, body_seq);
ADD_LABEL(ret, endlabel);
return COMPILE_OK;
}
static int
iseq_compile_pattern_each(rb_iseq_t *iseq, LINK_ANCHOR *const ret, const NODE *const node, int in_alt_pattern)
{
const int line = nd_line(node);
switch (nd_type(node)) {
case NODE_ARYPTN: {
/*
* if pattern.use_rest_num?
* rest_num = 0
* end
* if pattern.has_constant_node?
* unless pattern.constant === obj
* goto match_failed
* end
* end
* unless obj.respond_to?(:deconstruct)
* goto match_failed
* end
* d = obj.deconstruct
* unless Array === d
* goto type_error
* end
* min_argc = pattern.pre_args_num + pattern.post_args_num
* if pattern.has_rest_arg?
* unless d.length >= min_argc
* goto match_failed
* end
* else
* unless d.length == min_argc
* goto match_failed
* end
* end
* pattern.pre_args_num.each do |i|
* unless pattern.pre_args[i].match?(d[i])
* goto match_failed
* end
* end
* if pattern.use_rest_num?
* rest_num = d.length - min_argc
* if pattern.has_rest_arg? && pattern.has_rest_arg_id # not `*`, but `*rest`
* unless pattern.rest_arg.match?(d[pattern.pre_args_num, rest_num])
* goto match_failed
* end
* end
* end
* pattern.post_args_num.each do |i|
* j = pattern.pre_args_num + i
* j += rest_num
* unless pattern.post_args[i].match?(d[j])
* goto match_failed
* end
* end
* true
* goto fin
* type_error:
* FrozenCore.raise TypeError
* match_failed:
* false
* fin:
*/
struct rb_ary_pattern_info *apinfo = node->nd_apinfo;
const NODE *args = apinfo->pre_args;
const int pre_args_num = apinfo->pre_args ? rb_long2int(apinfo->pre_args->nd_alen) : 0;
const int post_args_num = apinfo->post_args ? rb_long2int(apinfo->post_args->nd_alen) : 0;
const int min_argc = pre_args_num + post_args_num;
const int use_rest_num = apinfo->rest_arg && (NODE_NAMED_REST_P(apinfo->rest_arg) ||
(!NODE_NAMED_REST_P(apinfo->rest_arg) && post_args_num > 0));
LABEL *match_failed, *type_error, *fin;
int i;
match_failed = NEW_LABEL(line);
type_error = NEW_LABEL(line);
fin = NEW_LABEL(line);
if (use_rest_num) {
ADD_INSN1(ret, line, putobject, INT2FIX(0)); /* allocate stack for rest_num */
ADD_INSN(ret, line, swap);
}
if (node->nd_pconst) {
ADD_INSN(ret, line, dup);
CHECK(COMPILE(ret, "constant", node->nd_pconst));
ADD_INSN1(ret, line, checkmatch, INT2FIX(VM_CHECKMATCH_TYPE_CASE));
ADD_INSNL(ret, line, branchunless, match_failed);
}
ADD_INSN(ret, line, dup);
ADD_INSN1(ret, line, putobject, ID2SYM(rb_intern("deconstruct")));
ADD_SEND(ret, line, idRespond_to, INT2FIX(1));
ADD_INSNL(ret, line, branchunless, match_failed);
ADD_SEND(ret, line, rb_intern("deconstruct"), INT2FIX(0));
ADD_INSN(ret, line, dup);
2019-06-10 18:20:10 +03:00
ADD_INSN1(ret, line, checktype, INT2FIX(T_ARRAY));
ADD_INSNL(ret, line, branchunless, type_error);
ADD_INSN(ret, line, dup);
ADD_SEND(ret, line, idLength, INT2FIX(0));
ADD_INSN1(ret, line, putobject, INT2FIX(min_argc));
ADD_SEND(ret, line, apinfo->rest_arg ? idGE : idEq, INT2FIX(1));
ADD_INSNL(ret, line, branchunless, match_failed);
for (i = 0; i < pre_args_num; i++) {
ADD_INSN(ret, line, dup);
ADD_INSN1(ret, line, putobject, INT2FIX(i));
ADD_SEND(ret, line, idAREF, INT2FIX(1));
iseq_compile_pattern_each(iseq, ret, args->nd_head, in_alt_pattern);
args = args->nd_next;
ADD_INSNL(ret, line, branchunless, match_failed);
}
if (apinfo->rest_arg) {
if (NODE_NAMED_REST_P(apinfo->rest_arg)) {
ADD_INSN(ret, line, dup);
ADD_INSN1(ret, line, putobject, INT2FIX(pre_args_num));
ADD_INSN1(ret, line, topn, INT2FIX(1));
ADD_SEND(ret, line, idLength, INT2FIX(0));
ADD_INSN1(ret, line, putobject, INT2FIX(min_argc));
ADD_SEND(ret, line, idMINUS, INT2FIX(1));
ADD_INSN1(ret, line, setn, INT2FIX(4));
ADD_SEND(ret, line, idAREF, INT2FIX(2));
iseq_compile_pattern_each(iseq, ret, apinfo->rest_arg, in_alt_pattern);
ADD_INSNL(ret, line, branchunless, match_failed);
}
else {
if (post_args_num > 0) {
ADD_INSN(ret, line, dup);
ADD_SEND(ret, line, idLength, INT2FIX(0));
ADD_INSN1(ret, line, putobject, INT2FIX(min_argc));
ADD_SEND(ret, line, idMINUS, INT2FIX(1));
ADD_INSN1(ret, line, setn, INT2FIX(2));
ADD_INSN(ret, line, pop);
}
}
}
args = apinfo->post_args;
for (i = 0; i < post_args_num; i++) {
ADD_INSN(ret, line, dup);
ADD_INSN1(ret, line, putobject, INT2FIX(pre_args_num + i));
ADD_INSN1(ret, line, topn, INT2FIX(3));
ADD_SEND(ret, line, idPLUS, INT2FIX(1));
ADD_SEND(ret, line, idAREF, INT2FIX(1));
iseq_compile_pattern_each(iseq, ret, args->nd_head, in_alt_pattern);
args = args->nd_next;
ADD_INSNL(ret, line, branchunless, match_failed);
}
ADD_INSN(ret, line, pop);
if (use_rest_num) {
ADD_INSN(ret, line, pop);
}
ADD_INSN1(ret, line, putobject, Qtrue);
ADD_INSNL(ret, line, jump, fin);
ADD_LABEL(ret, type_error);
ADD_INSN1(ret, line, putspecialobject, INT2FIX(VM_SPECIAL_OBJECT_VMCORE));
ADD_INSN1(ret, line, putobject, rb_eTypeError);
ADD_INSN1(ret, line, putobject, rb_fstring_lit("deconstruct must return Array"));
ADD_SEND(ret, line, id_core_raise, INT2FIX(2));
ADD_LABEL(ret, match_failed);
ADD_INSN(ret, line, pop);
if (use_rest_num) {
ADD_INSN(ret, line, pop);
}
ADD_INSN1(ret, line, putobject, Qfalse);
ADD_LABEL(ret, fin);
break;
}
case NODE_HSHPTN: {
/*
* keys = nil
* if pattern.has_kw_args_node? && !pattern.has_kw_rest_arg_node?
* keys = pattern.kw_args_node.keys
* end
* if pattern.has_constant_node?
* unless pattern.constant === obj
* goto match_failed
* end
* end
* unless obj.respond_to?(:deconstruct_keys)
* goto match_failed
* end
* d = obj.deconstruct_keys(keys)
* unless Hash === d
* goto type_error
* end
* if pattern.has_kw_rest_arg_node?
* d = d.dup
* end
* if pattern.has_kw_args_node?
* pattern.kw_args_node.each |k,|
* unless d.key?(k)
* goto match_failed
* end
* end
* pattern.kw_args_node.each |k, pat|
* if pattern.has_kw_rest_arg_node?
* unless pat.match?(d.delete(k))
* goto match_failed
* end
* else
* unless pat.match?(d[k])
* goto match_failed
* end
* end
* end
* else
* unless d.empty?
* goto match_failed
* end
* end
* if pattern.has_kw_rest_arg_node?
* if pattern.no_rest_keyword?
* unless d.empty?
* goto match_failed
* end
* else
* unless pattern.kw_rest_arg_node.match?(d)
* goto match_failed
* end
* end
* end
* true
* goto fin
* type_error:
* FrozenCore.raise TypeError
* match_failed:
* false
* fin:
*/
LABEL *match_failed, *type_error, *fin;
VALUE keys = Qnil;
match_failed = NEW_LABEL(line);
type_error = NEW_LABEL(line);
fin = NEW_LABEL(line);
if (node->nd_pkwargs && !node->nd_pkwrestarg) {
const NODE *kw_args = node->nd_pkwargs->nd_head;
keys = rb_ary_new_capa(kw_args ? kw_args->nd_alen/2 : 0);
while (kw_args) {
rb_ary_push(keys, kw_args->nd_head->nd_lit);
kw_args = kw_args->nd_next->nd_next;
}
}
if (node->nd_pconst) {
ADD_INSN(ret, line, dup);
CHECK(COMPILE(ret, "constant", node->nd_pconst));
ADD_INSN1(ret, line, checkmatch, INT2FIX(VM_CHECKMATCH_TYPE_CASE));
ADD_INSNL(ret, line, branchunless, match_failed);
}
ADD_INSN(ret, line, dup);
ADD_INSN1(ret, line, putobject, ID2SYM(rb_intern("deconstruct_keys")));
ADD_SEND(ret, line, idRespond_to, INT2FIX(1));
ADD_INSNL(ret, line, branchunless, match_failed);
if (NIL_P(keys)) {
ADD_INSN(ret, line, putnil);
}
else {
ADD_INSN1(ret, line, duparray, keys);
RB_OBJ_WRITTEN(iseq, Qundef, rb_obj_hide(keys));
}
ADD_SEND(ret, line, rb_intern("deconstruct_keys"), INT2FIX(1));
ADD_INSN(ret, line, dup);
2019-06-10 18:20:10 +03:00
ADD_INSN1(ret, line, checktype, INT2FIX(T_HASH));
ADD_INSNL(ret, line, branchunless, type_error);
if (node->nd_pkwrestarg) {
ADD_SEND(ret, line, rb_intern("dup"), INT2FIX(0));
}
if (node->nd_pkwargs) {
int i;
int keys_num;
const NODE *args;
args = node->nd_pkwargs->nd_head;
if (args) {
DECL_ANCHOR(match_values);
INIT_ANCHOR(match_values);
keys_num = rb_long2int(args->nd_alen) / 2;
for (i = 0; i < keys_num; i++) {
NODE *key_node = args->nd_head;
NODE *value_node = args->nd_next->nd_head;
VALUE key;
if (nd_type(key_node) != NODE_LIT) {
UNKNOWN_NODE("NODE_IN", key_node, COMPILE_NG);
}
key = key_node->nd_lit;
ADD_INSN(ret, line, dup);
ADD_INSN1(ret, line, putobject, key);
ADD_SEND(ret, line, rb_intern("key?"), INT2FIX(1));
ADD_INSNL(ret, line, branchunless, match_failed);
ADD_INSN(match_values, line, dup);
ADD_INSN1(match_values, line, putobject, key);
ADD_SEND(match_values, line, node->nd_pkwrestarg ? rb_intern("delete") : idAREF, INT2FIX(1));
iseq_compile_pattern_each(iseq, match_values, value_node, in_alt_pattern);
ADD_INSNL(match_values, line, branchunless, match_failed);
args = args->nd_next->nd_next;
}
ADD_SEQ(ret, match_values);
}
}
else {
ADD_INSN(ret, line, dup);
ADD_SEND(ret, line, idEmptyP, INT2FIX(0));
ADD_INSNL(ret, line, branchunless, match_failed);
}
if (node->nd_pkwrestarg) {
if (node->nd_pkwrestarg == NODE_SPECIAL_NO_REST_KEYWORD) {
ADD_INSN(ret, line, dup);
ADD_SEND(ret, line, idEmptyP, INT2FIX(0));
ADD_INSNL(ret, line, branchunless, match_failed);
}
else {
ADD_INSN(ret, line, dup);
iseq_compile_pattern_each(iseq, ret, node->nd_pkwrestarg, in_alt_pattern);
ADD_INSNL(ret, line, branchunless, match_failed);
}
}
ADD_INSN(ret, line, pop);
ADD_INSN1(ret, line, putobject, Qtrue);
ADD_INSNL(ret, line, jump, fin);
ADD_LABEL(ret, type_error);
ADD_INSN1(ret, line, putspecialobject, INT2FIX(VM_SPECIAL_OBJECT_VMCORE));
ADD_INSN1(ret, line, putobject, rb_eTypeError);
ADD_INSN1(ret, line, putobject, rb_fstring_lit("deconstruct_keys must return Hash"));
ADD_SEND(ret, line, id_core_raise, INT2FIX(2));
ADD_LABEL(ret, match_failed);
ADD_INSN(ret, line, pop);
ADD_INSN1(ret, line, putobject, Qfalse);
ADD_LABEL(ret, fin);
break;
}
case NODE_LIT:
case NODE_STR:
case NODE_XSTR:
case NODE_DSTR:
case NODE_DSYM:
case NODE_DREGX:
case NODE_LIST:
case NODE_ZLIST:
case NODE_LAMBDA:
case NODE_DOT2:
case NODE_DOT3:
case NODE_CONST:
case NODE_LVAR:
case NODE_DVAR:
case NODE_TRUE:
case NODE_FALSE:
case NODE_SELF:
case NODE_NIL:
case NODE_COLON2:
case NODE_COLON3:
CHECK(COMPILE(ret, "case in literal", node));
ADD_INSN1(ret, line, checkmatch, INT2FIX(VM_CHECKMATCH_TYPE_CASE));
break;
case NODE_LASGN: {
struct rb_iseq_constant_body *const body = iseq->body;
ID id = node->nd_vid;
int idx = body->local_iseq->body->local_table_size - get_local_var_idx(iseq, id);
if (in_alt_pattern) {
const char *name = rb_id2name(id);
if (name && strlen(name) > 0 && name[0] != '_') {
COMPILE_ERROR(ERROR_ARGS "illegal variable in alternative pattern (%"PRIsVALUE")",
rb_id2str(id));
return COMPILE_NG;
}
}
ADD_SETLOCAL(ret, line, idx, get_lvar_level(iseq));
ADD_INSN1(ret, line, putobject, Qtrue);
break;
}
case NODE_DASGN:
case NODE_DASGN_CURR: {
int idx, lv, ls;
ID id = node->nd_vid;
idx = get_dyna_var_idx(iseq, id, &lv, &ls);
if (in_alt_pattern) {
const char *name = rb_id2name(id);
if (name && strlen(name) > 0 && name[0] != '_') {
COMPILE_ERROR(ERROR_ARGS "illegal variable in alternative pattern (%"PRIsVALUE")",
rb_id2str(id));
return COMPILE_NG;
}
}
if (idx < 0) {
COMPILE_ERROR(ERROR_ARGS "NODE_DASGN(_CURR): unknown id (%"PRIsVALUE")",
rb_id2str(id));
return COMPILE_NG;
}
ADD_SETLOCAL(ret, line, ls - idx, lv);
ADD_INSN1(ret, line, putobject, Qtrue);
break;
}
case NODE_IF:
case NODE_UNLESS: {
LABEL *match_failed, *fin;
match_failed = NEW_LABEL(line);
fin = NEW_LABEL(line);
iseq_compile_pattern_each(iseq, ret, node->nd_body, in_alt_pattern);
ADD_INSNL(ret, line, branchunless, match_failed);
CHECK(COMPILE(ret, "case in if", node->nd_cond));
if (nd_type(node) == NODE_IF) {
ADD_INSNL(ret, line, branchunless, match_failed);
}
else {
ADD_INSNL(ret, line, branchif, match_failed);
}
ADD_INSN1(ret, line, putobject, Qtrue);
ADD_INSNL(ret, line, jump, fin);
ADD_LABEL(ret, match_failed);
ADD_INSN1(ret, line, putobject, Qfalse);
ADD_LABEL(ret, fin);
break;
}
case NODE_HASH: {
NODE *n;
LABEL *match_failed, *fin;
match_failed = NEW_LABEL(line);
fin = NEW_LABEL(line);
n = node->nd_head;
if (! (nd_type(n) == NODE_LIST && n->nd_alen == 2)) {
COMPILE_ERROR(ERROR_ARGS "unexpected node");
return COMPILE_NG;
}
ADD_INSN(ret, line, dup);
iseq_compile_pattern_each(iseq, ret, n->nd_head, in_alt_pattern);
ADD_INSNL(ret, line, branchunless, match_failed);
iseq_compile_pattern_each(iseq, ret, n->nd_next->nd_head, in_alt_pattern);
ADD_INSNL(ret, line, jump, fin);
ADD_LABEL(ret, match_failed);
ADD_INSN(ret, line, pop);
ADD_INSN1(ret, line, putobject, Qfalse);
ADD_LABEL(ret, fin);
break;
}
case NODE_OR: {
LABEL *match_succeeded, *fin;
match_succeeded = NEW_LABEL(line);
fin = NEW_LABEL(line);
ADD_INSN(ret, line, dup);
iseq_compile_pattern_each(iseq, ret, node->nd_1st, TRUE);
ADD_INSNL(ret, line, branchif, match_succeeded);
iseq_compile_pattern_each(iseq, ret, node->nd_2nd, TRUE);
ADD_INSNL(ret, line, jump, fin);
ADD_LABEL(ret, match_succeeded);
ADD_INSN(ret, line, pop);
ADD_INSN1(ret, line, putobject, Qtrue);
ADD_LABEL(ret, fin);
break;
}
default:
UNKNOWN_NODE("NODE_IN", node, COMPILE_NG);
}
return COMPILE_OK;
}
static int
compile_case3(rb_iseq_t *iseq, LINK_ANCHOR *const ret, const NODE *const orig_node, int popped)
{
const NODE *pattern;
const NODE *node = orig_node;
LABEL *endlabel, *elselabel;
DECL_ANCHOR(head);
DECL_ANCHOR(body_seq);
DECL_ANCHOR(cond_seq);
int line, lineno, column, last_lineno, last_column;
enum node_type type;
VALUE branches = 0;
INIT_ANCHOR(head);
INIT_ANCHOR(body_seq);
INIT_ANCHOR(cond_seq);
CHECK(COMPILE(head, "case base", node->nd_head));
DECL_BRANCH_BASE(branches, nd_first_lineno(node), nd_first_column(node), nd_last_lineno(node), nd_last_column(node), "case");
node = node->nd_body;
EXPECT_NODE("NODE_CASE3", node, NODE_IN, COMPILE_NG);
type = nd_type(node);
line = nd_line(node);
lineno = nd_first_lineno(node);
column = nd_first_column(node);
last_lineno = nd_last_lineno(node);
last_column = nd_last_column(node);
endlabel = NEW_LABEL(line);
elselabel = NEW_LABEL(line);
ADD_SEQ(ret, head); /* case VAL */
while (type == NODE_IN) {
LABEL *l1;
l1 = NEW_LABEL(line);
ADD_LABEL(body_seq, l1);
ADD_INSN(body_seq, line, pop);
ADD_TRACE_BRANCH_COVERAGE(
body_seq,
node->nd_body ? nd_first_lineno(node->nd_body) : lineno,
node->nd_body ? nd_first_column(node->nd_body) : column,
node->nd_body ? nd_last_lineno(node->nd_body) : last_lineno,
node->nd_body ? nd_last_column(node->nd_body) : last_column,
"in",
branches);
CHECK(COMPILE_(body_seq, "in body", node->nd_body, popped));
ADD_INSNL(body_seq, line, jump, endlabel);
pattern = node->nd_head;
if (pattern) {
ADD_INSN (cond_seq, nd_line(pattern), dup);
iseq_compile_pattern_each(iseq, cond_seq, pattern, FALSE);
ADD_INSNL(cond_seq, nd_line(pattern), branchif, l1);
}
else {
COMPILE_ERROR(ERROR_ARGS "unexpected node");
return COMPILE_NG;
}
node = node->nd_next;
if (!node) {
break;
}
type = nd_type(node);
line = nd_line(node);
lineno = nd_first_lineno(node);
column = nd_first_column(node);
last_lineno = nd_last_lineno(node);
last_column = nd_last_column(node);
}
/* else */
if (node) {
ADD_LABEL(cond_seq, elselabel);
ADD_INSN(cond_seq, line, pop);
ADD_TRACE_BRANCH_COVERAGE(cond_seq, nd_first_lineno(node), nd_first_column(node), nd_last_lineno(node), nd_last_column(node), "else", branches);
CHECK(COMPILE_(cond_seq, "else", node, popped));
ADD_INSNL(cond_seq, line, jump, endlabel);
}
else {
debugs("== else (implicit)\n");
ADD_LABEL(cond_seq, elselabel);
ADD_TRACE_BRANCH_COVERAGE(cond_seq, nd_first_lineno(orig_node), nd_first_column(orig_node), nd_last_lineno(orig_node), nd_last_column(orig_node), "else", branches);
ADD_INSN1(cond_seq, nd_line(orig_node), putspecialobject, INT2FIX(VM_SPECIAL_OBJECT_VMCORE));
ADD_INSN1(cond_seq, nd_line(orig_node), putobject, rb_eNoMatchingPatternError);
ADD_INSN1(cond_seq, nd_line(orig_node), topn, INT2FIX(2));
ADD_SEND(cond_seq, nd_line(orig_node), id_core_raise, INT2FIX(2));
ADD_INSN(cond_seq, nd_line(orig_node), pop);
ADD_INSN(cond_seq, nd_line(orig_node), pop);
if (!popped) {
ADD_INSN(cond_seq, nd_line(orig_node), putnil);
}
ADD_INSNL(cond_seq, nd_line(orig_node), jump, endlabel);
}
ADD_SEQ(ret, cond_seq);
ADD_SEQ(ret, body_seq);
ADD_LABEL(ret, endlabel);
return COMPILE_OK;
}
static int
compile_loop(rb_iseq_t *iseq, LINK_ANCHOR *const ret, const NODE *const node, int popped, const enum node_type type)
{
const int line = (int)nd_line(node);
const int lineno = nd_first_lineno(node);
const int column = nd_first_column(node);
const int last_lineno = nd_last_lineno(node);
const int last_column = nd_last_column(node);
LABEL *prev_start_label = ISEQ_COMPILE_DATA(iseq)->start_label;
LABEL *prev_end_label = ISEQ_COMPILE_DATA(iseq)->end_label;
LABEL *prev_redo_label = ISEQ_COMPILE_DATA(iseq)->redo_label;
int prev_loopval_popped = ISEQ_COMPILE_DATA(iseq)->loopval_popped;
VALUE branches = Qfalse;
struct iseq_compile_data_ensure_node_stack enl;
LABEL *next_label = ISEQ_COMPILE_DATA(iseq)->start_label = NEW_LABEL(line); /* next */
LABEL *redo_label = ISEQ_COMPILE_DATA(iseq)->redo_label = NEW_LABEL(line); /* redo */
LABEL *break_label = ISEQ_COMPILE_DATA(iseq)->end_label = NEW_LABEL(line); /* break */
LABEL *end_label = NEW_LABEL(line);
LABEL *adjust_label = NEW_LABEL(line);
LABEL *next_catch_label = NEW_LABEL(line);
LABEL *tmp_label = NULL;
ISEQ_COMPILE_DATA(iseq)->loopval_popped = 0;
push_ensure_entry(iseq, &enl, NULL, NULL);
if (node->nd_state == 1) {
ADD_INSNL(ret, line, jump, next_label);
}
else {
tmp_label = NEW_LABEL(line);
ADD_INSNL(ret, line, jump, tmp_label);
}
ADD_LABEL(ret, adjust_label);
ADD_INSN(ret, line, putnil);
ADD_LABEL(ret, next_catch_label);
ADD_INSN(ret, line, pop);
ADD_INSNL(ret, line, jump, next_label);
if (tmp_label) ADD_LABEL(ret, tmp_label);
ADD_LABEL(ret, redo_label);
DECL_BRANCH_BASE(branches, lineno, column, last_lineno, last_column, type == NODE_WHILE ? "while" : "until");
ADD_TRACE_BRANCH_COVERAGE(
ret,
node->nd_body ? nd_first_lineno(node->nd_body) : lineno,
node->nd_body ? nd_first_column(node->nd_body) : column,
node->nd_body ? nd_last_lineno(node->nd_body) : last_lineno,
node->nd_body ? nd_last_column(node->nd_body) : last_column,
"body",
branches);
CHECK(COMPILE_POPPED(ret, "while body", node->nd_body));
ADD_LABEL(ret, next_label); /* next */
if (type == NODE_WHILE) {
compile_branch_condition(iseq, ret, node->nd_cond,
redo_label, end_label);
}
else {
/* until */
compile_branch_condition(iseq, ret, node->nd_cond,
end_label, redo_label);
}
ADD_LABEL(ret, end_label);
ADD_ADJUST_RESTORE(ret, adjust_label);
if (node->nd_state == Qundef) {
/* ADD_INSN(ret, line, putundef); */
COMPILE_ERROR(ERROR_ARGS "unsupported: putundef");
return COMPILE_NG;
}
else {
ADD_INSN(ret, line, putnil);
}
ADD_LABEL(ret, break_label); /* break */
if (popped) {
ADD_INSN(ret, line, pop);
}
ADD_CATCH_ENTRY(CATCH_TYPE_BREAK, redo_label, break_label, NULL,
break_label);
ADD_CATCH_ENTRY(CATCH_TYPE_NEXT, redo_label, break_label, NULL,
next_catch_label);
ADD_CATCH_ENTRY(CATCH_TYPE_REDO, redo_label, break_label, NULL,
ISEQ_COMPILE_DATA(iseq)->redo_label);
ISEQ_COMPILE_DATA(iseq)->start_label = prev_start_label;
ISEQ_COMPILE_DATA(iseq)->end_label = prev_end_label;
ISEQ_COMPILE_DATA(iseq)->redo_label = prev_redo_label;
ISEQ_COMPILE_DATA(iseq)->loopval_popped = prev_loopval_popped;
ISEQ_COMPILE_DATA(iseq)->ensure_node_stack = ISEQ_COMPILE_DATA(iseq)->ensure_node_stack->prev;
return COMPILE_OK;
}
static int
compile_iter(rb_iseq_t *iseq, LINK_ANCHOR *const ret, const NODE *const node, int popped)
{
const int line = nd_line(node);
const rb_iseq_t *prevblock = ISEQ_COMPILE_DATA(iseq)->current_block;
LABEL *retry_label = NEW_LABEL(line);
LABEL *retry_end_l = NEW_LABEL(line);
const rb_iseq_t *child_iseq;
ADD_LABEL(ret, retry_label);
if (nd_type(node) == NODE_FOR) {
CHECK(COMPILE(ret, "iter caller (for)", node->nd_iter));
ISEQ_COMPILE_DATA(iseq)->current_block = child_iseq =
NEW_CHILD_ISEQ(node->nd_body, make_name_for_block(iseq),
ISEQ_TYPE_BLOCK, line);
ADD_SEND_WITH_BLOCK(ret, line, idEach, INT2FIX(0), child_iseq);
}
else {
ISEQ_COMPILE_DATA(iseq)->current_block = child_iseq =
NEW_CHILD_ISEQ(node->nd_body, make_name_for_block(iseq),
ISEQ_TYPE_BLOCK, line);
CHECK(COMPILE(ret, "iter caller", node->nd_iter));
}
ADD_LABEL(ret, retry_end_l);
if (popped) {
ADD_INSN(ret, line, pop);
}
ISEQ_COMPILE_DATA(iseq)->current_block = prevblock;
ADD_CATCH_ENTRY(CATCH_TYPE_BREAK, retry_label, retry_end_l, child_iseq, retry_end_l);
return COMPILE_OK;
}
static int
compile_for_masgn(rb_iseq_t *iseq, LINK_ANCHOR *const ret, const NODE *const node, int popped)
{
/* massign to var in "for"
* (args.length == 1 && Array.try_convert(args[0])) || args
*/
const int line = nd_line(node);
const NODE *var = node->nd_var;
LABEL *not_single = NEW_LABEL(nd_line(var));
LABEL *not_ary = NEW_LABEL(nd_line(var));
CHECK(COMPILE(ret, "for var", var));
ADD_INSN(ret, line, dup);
ADD_CALL(ret, line, idLength, INT2FIX(0));
ADD_INSN1(ret, line, putobject, INT2FIX(1));
ADD_CALL(ret, line, idEq, INT2FIX(1));
ADD_INSNL(ret, line, branchunless, not_single);
ADD_INSN(ret, line, dup);
ADD_INSN1(ret, line, putobject, INT2FIX(0));
ADD_CALL(ret, line, idAREF, INT2FIX(1));
ADD_INSN1(ret, line, putobject, rb_cArray);
ADD_INSN(ret, line, swap);
ADD_CALL(ret, line, rb_intern("try_convert"), INT2FIX(1));
ADD_INSN(ret, line, dup);
ADD_INSNL(ret, line, branchunless, not_ary);
ADD_INSN(ret, line, swap);
ADD_LABEL(ret, not_ary);
ADD_INSN(ret, line, pop);
ADD_LABEL(ret, not_single);
return COMPILE_OK;
}
static int
compile_break(rb_iseq_t *iseq, LINK_ANCHOR *const ret, const NODE *const node, int popped)
{
const int line = nd_line(node);
unsigned long throw_flag = 0;
if (ISEQ_COMPILE_DATA(iseq)->redo_label != 0) {
/* while/until */
LABEL *splabel = NEW_LABEL(0);
ADD_LABEL(ret, splabel);
ADD_ADJUST(ret, line, ISEQ_COMPILE_DATA(iseq)->redo_label);
CHECK(COMPILE_(ret, "break val (while/until)", node->nd_stts,
ISEQ_COMPILE_DATA(iseq)->loopval_popped));
add_ensure_iseq(ret, iseq, 0);
ADD_INSNL(ret, line, jump, ISEQ_COMPILE_DATA(iseq)->end_label);
ADD_ADJUST_RESTORE(ret, splabel);
if (!popped) {
ADD_INSN(ret, line, putnil);
}
}
else if (iseq->body->type == ISEQ_TYPE_BLOCK) {
break_by_insn:
/* escape from block */
CHECK(COMPILE(ret, "break val (block)", node->nd_stts));
ADD_INSN1(ret, line, throw, INT2FIX(throw_flag | TAG_BREAK));
if (popped) {
ADD_INSN(ret, line, pop);
}
}
else if (iseq->body->type == ISEQ_TYPE_EVAL) {
break_in_eval:
COMPILE_ERROR(ERROR_ARGS "Can't escape from eval with break");
return COMPILE_NG;
}
else {
const rb_iseq_t *ip = iseq->body->parent_iseq;
while (ip) {
if (!ISEQ_COMPILE_DATA(ip)) {
ip = 0;
break;
}
if (ISEQ_COMPILE_DATA(ip)->redo_label != 0) {
throw_flag = VM_THROW_NO_ESCAPE_FLAG;
goto break_by_insn;
}
else if (ip->body->type == ISEQ_TYPE_BLOCK) {
goto break_by_insn;
}
else if (ip->body->type == ISEQ_TYPE_EVAL) {
goto break_in_eval;
}
ip = ip->body->parent_iseq;
}
COMPILE_ERROR(ERROR_ARGS "Invalid break");
return COMPILE_NG;
}
return COMPILE_OK;
}
static int
compile_next(rb_iseq_t *iseq, LINK_ANCHOR *const ret, const NODE *const node, int popped)
{
const int line = nd_line(node);
unsigned long throw_flag = 0;
if (ISEQ_COMPILE_DATA(iseq)->redo_label != 0) {
LABEL *splabel = NEW_LABEL(0);
debugs("next in while loop\n");
ADD_LABEL(ret, splabel);
CHECK(COMPILE(ret, "next val/valid syntax?", node->nd_stts));
add_ensure_iseq(ret, iseq, 0);
ADD_ADJUST(ret, line, ISEQ_COMPILE_DATA(iseq)->redo_label);
ADD_INSNL(ret, line, jump, ISEQ_COMPILE_DATA(iseq)->start_label);
ADD_ADJUST_RESTORE(ret, splabel);
if (!popped) {
ADD_INSN(ret, line, putnil);
}
}
else if (ISEQ_COMPILE_DATA(iseq)->end_label) {
LABEL *splabel = NEW_LABEL(0);
debugs("next in block\n");
ADD_LABEL(ret, splabel);
ADD_ADJUST(ret, line, ISEQ_COMPILE_DATA(iseq)->start_label);
CHECK(COMPILE(ret, "next val", node->nd_stts));
add_ensure_iseq(ret, iseq, 0);
ADD_INSNL(ret, line, jump, ISEQ_COMPILE_DATA(iseq)->end_label);
ADD_ADJUST_RESTORE(ret, splabel);
splabel->unremovable = FALSE;
if (!popped) {
ADD_INSN(ret, line, putnil);
}
}
else if (iseq->body->type == ISEQ_TYPE_EVAL) {
next_in_eval:
COMPILE_ERROR(ERROR_ARGS "Can't escape from eval with next");
return COMPILE_NG;
}
else {
const rb_iseq_t *ip = iseq;
while (ip) {
if (!ISEQ_COMPILE_DATA(ip)) {
ip = 0;
break;
}
throw_flag = VM_THROW_NO_ESCAPE_FLAG;
if (ISEQ_COMPILE_DATA(ip)->redo_label != 0) {
/* while loop */
break;
}
else if (ip->body->type == ISEQ_TYPE_BLOCK) {
break;
}
else if (ip->body->type == ISEQ_TYPE_EVAL) {
goto next_in_eval;
}
ip = ip->body->parent_iseq;
}
if (ip != 0) {
CHECK(COMPILE(ret, "next val", node->nd_stts));
ADD_INSN1(ret, line, throw, INT2FIX(throw_flag | TAG_NEXT));
if (popped) {
ADD_INSN(ret, line, pop);
}
}
else {
COMPILE_ERROR(ERROR_ARGS "Invalid next");
return COMPILE_NG;
}
}
return COMPILE_OK;
}
static int
compile_redo(rb_iseq_t *iseq, LINK_ANCHOR *const ret, const NODE *const node, int popped)
{
const int line = nd_line(node);
if (ISEQ_COMPILE_DATA(iseq)->redo_label) {
LABEL *splabel = NEW_LABEL(0);
debugs("redo in while");
ADD_LABEL(ret, splabel);
ADD_ADJUST(ret, line, ISEQ_COMPILE_DATA(iseq)->redo_label);
add_ensure_iseq(ret, iseq, 0);
ADD_INSNL(ret, line, jump, ISEQ_COMPILE_DATA(iseq)->redo_label);
ADD_ADJUST_RESTORE(ret, splabel);
if (!popped) {
ADD_INSN(ret, line, putnil);
}
}
else if (iseq->body->type == ISEQ_TYPE_EVAL) {
redo_in_eval:
COMPILE_ERROR(ERROR_ARGS "Can't escape from eval with redo");
return COMPILE_NG;
}
else if (ISEQ_COMPILE_DATA(iseq)->start_label) {
LABEL *splabel = NEW_LABEL(0);
debugs("redo in block");
ADD_LABEL(ret, splabel);
add_ensure_iseq(ret, iseq, 0);
ADD_ADJUST(ret, line, ISEQ_COMPILE_DATA(iseq)->start_label);
ADD_INSNL(ret, line, jump, ISEQ_COMPILE_DATA(iseq)->start_label);
ADD_ADJUST_RESTORE(ret, splabel);
if (!popped) {
ADD_INSN(ret, line, putnil);
}
}
else {
const rb_iseq_t *ip = iseq;
while (ip) {
if (!ISEQ_COMPILE_DATA(ip)) {
ip = 0;
break;
}
if (ISEQ_COMPILE_DATA(ip)->redo_label != 0) {
break;
}
else if (ip->body->type == ISEQ_TYPE_BLOCK) {
break;
}
else if (ip->body->type == ISEQ_TYPE_EVAL) {
goto redo_in_eval;
}
ip = ip->body->parent_iseq;
}
if (ip != 0) {
ADD_INSN(ret, line, putnil);
ADD_INSN1(ret, line, throw, INT2FIX(VM_THROW_NO_ESCAPE_FLAG | TAG_REDO));
if (popped) {
ADD_INSN(ret, line, pop);
}
}
else {
COMPILE_ERROR(ERROR_ARGS "Invalid redo");
return COMPILE_NG;
}
}
return COMPILE_OK;
}
static int
compile_retry(rb_iseq_t *iseq, LINK_ANCHOR *const ret, const NODE *const node, int popped)
{
const int line = nd_line(node);
if (iseq->body->type == ISEQ_TYPE_RESCUE) {
ADD_INSN(ret, line, putnil);
ADD_INSN1(ret, line, throw, INT2FIX(TAG_RETRY));
if (popped) {
ADD_INSN(ret, line, pop);
}
}
else {
COMPILE_ERROR(ERROR_ARGS "Invalid retry");
return COMPILE_NG;
}
return COMPILE_OK;
}
static int
compile_rescue(rb_iseq_t *iseq, LINK_ANCHOR *const ret, const NODE *const node, int popped)
{
const int line = nd_line(node);
LABEL *lstart = NEW_LABEL(line);
LABEL *lend = NEW_LABEL(line);
LABEL *lcont = NEW_LABEL(line);
const rb_iseq_t *rescue = NEW_CHILD_ISEQ(node->nd_resq,
rb_str_concat(rb_str_new2("rescue in "), iseq->body->location.label),
ISEQ_TYPE_RESCUE, line);
lstart->rescued = LABEL_RESCUE_BEG;
lend->rescued = LABEL_RESCUE_END;
ADD_LABEL(ret, lstart);
CHECK(COMPILE(ret, "rescue head", node->nd_head));
ADD_LABEL(ret, lend);
if (node->nd_else) {
ADD_INSN(ret, line, pop);
CHECK(COMPILE(ret, "rescue else", node->nd_else));
}
ADD_INSN(ret, line, nop);
ADD_LABEL(ret, lcont);
if (popped) {
ADD_INSN(ret, line, pop);
}
/* register catch entry */
ADD_CATCH_ENTRY(CATCH_TYPE_RESCUE, lstart, lend, rescue, lcont);
ADD_CATCH_ENTRY(CATCH_TYPE_RETRY, lend, lcont, NULL, lstart);
return COMPILE_OK;
}
static int
compile_resbody(rb_iseq_t *iseq, LINK_ANCHOR *const ret, const NODE *const node, int popped)
{
const int line = nd_line(node);
const NODE *resq = node;
const NODE *narg;
LABEL *label_miss, *label_hit;
while (resq) {
label_miss = NEW_LABEL(line);
label_hit = NEW_LABEL(line);
narg = resq->nd_args;
if (narg) {
switch (nd_type(narg)) {
case NODE_LIST:
while (narg) {
ADD_GETLOCAL(ret, line, LVAR_ERRINFO, 0);
CHECK(COMPILE(ret, "rescue arg", narg->nd_head));
ADD_INSN1(ret, line, checkmatch, INT2FIX(VM_CHECKMATCH_TYPE_RESCUE));
ADD_INSNL(ret, line, branchif, label_hit);
narg = narg->nd_next;
}
break;
case NODE_SPLAT:
case NODE_ARGSCAT:
case NODE_ARGSPUSH:
ADD_GETLOCAL(ret, line, LVAR_ERRINFO, 0);
CHECK(COMPILE(ret, "rescue/cond splat", narg));
ADD_INSN1(ret, line, checkmatch, INT2FIX(VM_CHECKMATCH_TYPE_RESCUE | VM_CHECKMATCH_ARRAY));
ADD_INSNL(ret, line, branchif, label_hit);
break;
default:
UNKNOWN_NODE("NODE_RESBODY", narg, COMPILE_NG);
}
}
else {
ADD_GETLOCAL(ret, line, LVAR_ERRINFO, 0);
ADD_INSN1(ret, line, putobject, rb_eStandardError);
ADD_INSN1(ret, line, checkmatch, INT2FIX(VM_CHECKMATCH_TYPE_RESCUE));
ADD_INSNL(ret, line, branchif, label_hit);
}
ADD_INSNL(ret, line, jump, label_miss);
ADD_LABEL(ret, label_hit);
CHECK(COMPILE(ret, "resbody body", resq->nd_body));
if (ISEQ_COMPILE_DATA(iseq)->option->tailcall_optimization) {
ADD_INSN(ret, line, nop);
}
ADD_INSN(ret, line, leave);
ADD_LABEL(ret, label_miss);
resq = resq->nd_head;
}
return COMPILE_OK;
}
static int
compile_ensure(rb_iseq_t *iseq, LINK_ANCHOR *const ret, const NODE *const node, int popped)
{
const int line = nd_line(node);
DECL_ANCHOR(ensr);
const rb_iseq_t *ensure = NEW_CHILD_ISEQ(node->nd_ensr,
rb_str_concat(rb_str_new2 ("ensure in "), iseq->body->location.label),
ISEQ_TYPE_ENSURE, line);
LABEL *lstart = NEW_LABEL(line);
LABEL *lend = NEW_LABEL(line);
LABEL *lcont = NEW_LABEL(line);
LINK_ELEMENT *last;
int last_leave = 0;
struct ensure_range er;
struct iseq_compile_data_ensure_node_stack enl;
struct ensure_range *erange;
INIT_ANCHOR(ensr);
CHECK(COMPILE_POPPED(ensr, "ensure ensr", node->nd_ensr));
last = ensr->last;
last_leave = last && IS_INSN(last) && IS_INSN_ID(last, leave);
er.begin = lstart;
er.end = lend;
er.next = 0;
push_ensure_entry(iseq, &enl, &er, node->nd_ensr);
ADD_LABEL(ret, lstart);
CHECK(COMPILE_(ret, "ensure head", node->nd_head, (popped | last_leave)));
ADD_LABEL(ret, lend);
ADD_SEQ(ret, ensr);
if (!popped && last_leave) ADD_INSN(ret, line, putnil);
ADD_LABEL(ret, lcont);
if (last_leave) ADD_INSN(ret, line, pop);
erange = ISEQ_COMPILE_DATA(iseq)->ensure_node_stack->erange;
if (lstart->link.next != &lend->link) {
while (erange) {
ADD_CATCH_ENTRY(CATCH_TYPE_ENSURE, erange->begin, erange->end,
ensure, lcont);
erange = erange->next;
}
}
ISEQ_COMPILE_DATA(iseq)->ensure_node_stack = enl.prev;
return COMPILE_OK;
}
static int
compile_return(rb_iseq_t *iseq, LINK_ANCHOR *const ret, const NODE *const node, int popped)
{
const int line = nd_line(node);
if (iseq) {
enum iseq_type type = iseq->body->type;
const rb_iseq_t *is = iseq;
enum iseq_type t = type;
const NODE *retval = node->nd_stts;
LABEL *splabel = 0;
while (t == ISEQ_TYPE_RESCUE || t == ISEQ_TYPE_ENSURE) {
if (!(is = is->body->parent_iseq)) break;
t = is->body->type;
}
switch (t) {
case ISEQ_TYPE_TOP:
case ISEQ_TYPE_MAIN:
if (retval) {
rb_warn("argument of top-level return is ignored");
}
if (is == iseq) {
/* plain top-level, leave directly */
type = ISEQ_TYPE_METHOD;
}
break;
default:
break;
}
if (type == ISEQ_TYPE_METHOD) {
splabel = NEW_LABEL(0);
ADD_LABEL(ret, splabel);
ADD_ADJUST(ret, line, 0);
}
CHECK(COMPILE(ret, "return nd_stts (return val)", retval));
if (type == ISEQ_TYPE_METHOD) {
add_ensure_iseq(ret, iseq, 1);
ADD_TRACE(ret, RUBY_EVENT_RETURN);
ADD_INSN(ret, line, leave);
ADD_ADJUST_RESTORE(ret, splabel);
if (!popped) {
ADD_INSN(ret, line, putnil);
}
}
else {
ADD_INSN1(ret, line, throw, INT2FIX(TAG_RETURN));
if (popped) {
ADD_INSN(ret, line, pop);
}
}
}
return COMPILE_OK;
}
static int
compile_evstr(rb_iseq_t *iseq, LINK_ANCHOR *const ret, const NODE *const node, int popped)
{
CHECK(COMPILE_(ret, "nd_body", node, popped));
if (!popped && !all_string_result_p(node)) {
const int line = nd_line(node);
const unsigned int flag = VM_CALL_FCALL;
LABEL *isstr = NEW_LABEL(line);
ADD_INSN(ret, line, dup);
ADD_INSN1(ret, line, checktype, INT2FIX(T_STRING));
ADD_INSNL(ret, line, branchif, isstr);
ADD_INSN(ret, line, dup);
ADD_SEND_R(ret, line, idTo_s, INT2FIX(0), NULL, INT2FIX(flag), NULL);
ADD_INSN(ret, line, tostring);
ADD_LABEL(ret, isstr);
}
return COMPILE_OK;
}
static LABEL *
qcall_branch_start(rb_iseq_t *iseq, LINK_ANCHOR *const recv, VALUE *branches, const NODE *node, int line)
{
LABEL *else_label = NEW_LABEL(line);
const int first_lineno = nd_first_lineno(node), first_column = nd_first_column(node);
const int last_lineno = nd_last_lineno(node), last_column = nd_last_column(node);
VALUE br = 0;
DECL_BRANCH_BASE(br, first_lineno, first_column, last_lineno, last_column, "&.");
*branches = br;
ADD_INSN(recv, line, dup);
ADD_INSNL(recv, line, branchnil, else_label);
ADD_TRACE_BRANCH_COVERAGE(recv, first_lineno, first_column, last_lineno, last_column, "then", br);
return else_label;
}
static void
qcall_branch_end(rb_iseq_t *iseq, LINK_ANCHOR *const ret, LABEL *else_label, VALUE branches, const NODE *node, int line)
{
LABEL *end_label;
if (!else_label) return;
end_label = NEW_LABEL(line);
ADD_INSNL(ret, line, jump, end_label);
ADD_LABEL(ret, else_label);
ADD_TRACE_BRANCH_COVERAGE(ret, nd_first_lineno(node), nd_first_column(node), nd_last_lineno(node), nd_last_column(node),
"else", branches);
ADD_LABEL(ret, end_label);
}
static int
compile_call_precheck_freeze(rb_iseq_t *iseq, LINK_ANCHOR *const ret, const NODE *const node, int line, int popped)
{
/* optimization shortcut
* "literal".freeze -> opt_str_freeze("literal")
*/
if (node->nd_recv && nd_type(node->nd_recv) == NODE_STR &&
(node->nd_mid == idFreeze || node->nd_mid == idUMinus) &&
node->nd_args == NULL &&
ISEQ_COMPILE_DATA(iseq)->current_block == NULL &&
ISEQ_COMPILE_DATA(iseq)->option->specialized_instruction) {
VALUE str = rb_fstring(node->nd_recv->nd_lit);
if (node->nd_mid == idUMinus) {
ADD_INSN2(ret, line, opt_str_uminus, str,
new_callinfo(iseq, idUMinus, 0, 0, NULL, FALSE));
}
else {
ADD_INSN2(ret, line, opt_str_freeze, str,
new_callinfo(iseq, idFreeze, 0, 0, NULL, FALSE));
}
RB_OBJ_WRITTEN(iseq, Qundef, str);
if (popped) {
ADD_INSN(ret, line, pop);
}
return TRUE;
}
/* optimization shortcut
* obj["literal"] -> opt_aref_with(obj, "literal")
*/
if (node->nd_mid == idAREF && !private_recv_p(node) && node->nd_args &&
nd_type(node->nd_args) == NODE_LIST && node->nd_args->nd_alen == 1 &&
nd_type(node->nd_args->nd_head) == NODE_STR &&
ISEQ_COMPILE_DATA(iseq)->current_block == NULL &&
!ISEQ_COMPILE_DATA(iseq)->option->frozen_string_literal &&
ISEQ_COMPILE_DATA(iseq)->option->specialized_instruction) {
VALUE str = rb_fstring(node->nd_args->nd_head->nd_lit);
CHECK(COMPILE(ret, "recv", node->nd_recv));
ADD_INSN2(ret, line, opt_aref_with, str,
new_callinfo(iseq, idAREF, 1, 0, NULL, FALSE));
RB_OBJ_WRITTEN(iseq, Qundef, str);
if (popped) {
ADD_INSN(ret, line, pop);
}
return TRUE;
}
return FALSE;
}
static int
compile_call(rb_iseq_t *iseq, LINK_ANCHOR *const ret, const NODE *const node, int type, int line, int popped)
{
/* call: obj.method(...)
* fcall: func(...)
* vcall: func
*/
DECL_ANCHOR(recv);
DECL_ANCHOR(args);
ID mid = node->nd_mid;
VALUE argc;
unsigned int flag = 0;
struct rb_call_info_kw_arg *keywords = NULL;
const rb_iseq_t *parent_block = ISEQ_COMPILE_DATA(iseq)->current_block;
LABEL *else_label = NULL;
VALUE branches = Qfalse;
ISEQ_COMPILE_DATA(iseq)->current_block = NULL;
INIT_ANCHOR(recv);
INIT_ANCHOR(args);
#if OPT_SUPPORT_JOKE
if (nd_type(node) == NODE_VCALL) {
ID id_bitblt;
ID id_answer;
CONST_ID(id_bitblt, "bitblt");
CONST_ID(id_answer, "the_answer_to_life_the_universe_and_everything");
if (mid == id_bitblt) {
ADD_INSN(ret, line, bitblt);
return COMPILE_OK;
}
else if (mid == id_answer) {
ADD_INSN(ret, line, answer);
return COMPILE_OK;
}
}
/* only joke */
{
ID goto_id;
ID label_id;
CONST_ID(goto_id, "__goto__");
CONST_ID(label_id, "__label__");
if (nd_type(node) == NODE_FCALL &&
(mid == goto_id || mid == label_id)) {
LABEL *label;
st_data_t data;
st_table *labels_table = ISEQ_COMPILE_DATA(iseq)->labels_table;
VALUE label_name;
if (!labels_table) {
labels_table = st_init_numtable();
ISEQ_COMPILE_DATA(iseq)->labels_table = labels_table;
}
if (nd_type(node->nd_args->nd_head) == NODE_LIT &&
SYMBOL_P(node->nd_args->nd_head->nd_lit)) {
label_name = node->nd_args->nd_head->nd_lit;
if (!st_lookup(labels_table, (st_data_t)label_name, &data)) {
label = NEW_LABEL(line);
label->position = line;
st_insert(labels_table, (st_data_t)label_name, (st_data_t)label);
}
else {
label = (LABEL *)data;
}
}
else {
COMPILE_ERROR(ERROR_ARGS "invalid goto/label format");
return COMPILE_NG;
}
if (mid == goto_id) {
ADD_INSNL(ret, line, jump, label);
}
else {
ADD_LABEL(ret, label);
}
return COMPILE_OK;
}
}
#endif
/* receiver */
if (type == NODE_CALL || type == NODE_OPCALL || type == NODE_QCALL) {
int idx, level;
if (mid == idCall &&
nd_type(node->nd_recv) == NODE_LVAR &&
iseq_block_param_id_p(iseq, node->nd_recv->nd_vid, &idx, &level)) {
ADD_INSN2(recv, nd_line(node->nd_recv), getblockparamproxy, INT2FIX(idx + VM_ENV_DATA_SIZE - 1), INT2FIX(level));
}
else if (private_recv_p(node)) {
ADD_INSN(recv, nd_line(node), putself);
flag |= VM_CALL_FCALL;
}
else {
CHECK(COMPILE(recv, "recv", node->nd_recv));
}
if (type == NODE_QCALL) {
else_label = qcall_branch_start(iseq, recv, &branches, node, line);
}
}
else if (type == NODE_FCALL || type == NODE_VCALL) {
ADD_CALL_RECEIVER(recv, line);
}
/* args */
if (type != NODE_VCALL) {
argc = setup_args(iseq, args, node->nd_args, &flag, &keywords);
CHECK(!NIL_P(argc));
}
else {
argc = INT2FIX(0);
}
ADD_SEQ(ret, recv);
ADD_SEQ(ret, args);
debugp_param("call args argc", argc);
debugp_param("call method", ID2SYM(mid));
switch ((int)type) {
case NODE_VCALL:
flag |= VM_CALL_VCALL;
/* VCALL is funcall, so fall through */
case NODE_FCALL:
flag |= VM_CALL_FCALL;
}
ADD_SEND_R(ret, line, mid, argc, parent_block, INT2FIX(flag), keywords);
qcall_branch_end(iseq, ret, else_label, branches, node, line);
if (popped) {
ADD_INSN(ret, line, pop);
}
return COMPILE_OK;
}
static int iseq_compile_each0(rb_iseq_t *iseq, LINK_ANCHOR *const ret, const NODE *node, int popped);
/**
compile each node
self: InstructionSequence
node: Ruby compiled node
popped: This node will be popped
*/
static int
iseq_compile_each(rb_iseq_t *iseq, LINK_ANCHOR *ret, const NODE *node, int popped)
{
if (node == 0) {
if (!popped) {
int lineno = ISEQ_COMPILE_DATA(iseq)->last_line;
if (lineno == 0) lineno = FIX2INT(rb_iseq_first_lineno(iseq));
debugs("node: NODE_NIL(implicit)\n");
ADD_INSN(ret, lineno, putnil);
}
return COMPILE_OK;
}
return iseq_compile_each0(iseq, ret, node, popped);
}
static int
check_yield_place(const rb_iseq_t *iseq, int line)
{
VALUE file;
switch (iseq->body->local_iseq->body->type) {
case ISEQ_TYPE_TOP:
case ISEQ_TYPE_MAIN:
return FALSE;
case ISEQ_TYPE_CLASS:
file = rb_iseq_path(iseq);
rb_compile_warn(RSTRING_PTR(file), line,
"`yield' in class syntax will not be supported from Ruby 3.0. [Feature #15575]");
return TRUE;
default:
return TRUE;
}
}
static int
iseq_compile_each0(rb_iseq_t *iseq, LINK_ANCHOR *const ret, const NODE *node, int popped)
{
const int line = (int)nd_line(node);
const enum node_type type = nd_type(node);
struct rb_iseq_constant_body *const body = iseq->body;
if (ISEQ_COMPILE_DATA(iseq)->last_line == line) {
/* ignore */
}
else {
if (node->flags & NODE_FL_NEWLINE) {
int event = RUBY_EVENT_LINE;
ISEQ_COMPILE_DATA(iseq)->last_line = line;
if (ISEQ_COVERAGE(iseq) && ISEQ_LINE_COVERAGE(iseq)) {
event |= RUBY_EVENT_COVERAGE_LINE;
}
ADD_TRACE(ret, event);
}
}
debug_node_start(node);
#undef BEFORE_RETURN
#define BEFORE_RETURN debug_node_end()
switch (type) {
case NODE_BLOCK:{
while (node && nd_type(node) == NODE_BLOCK) {
CHECK(COMPILE_(ret, "BLOCK body", node->nd_head,
(node->nd_next ? 1 : popped)));
node = node->nd_next;
}
if (node) {
CHECK(COMPILE_(ret, "BLOCK next", node->nd_next, popped));
}
break;
}
case NODE_IF:
case NODE_UNLESS:
CHECK(compile_if(iseq, ret, node, popped, type));
break;
case NODE_CASE:
CHECK(compile_case(iseq, ret, node, popped));
break;
case NODE_CASE2:
CHECK(compile_case2(iseq, ret, node, popped));
break;
case NODE_CASE3:
CHECK(compile_case3(iseq, ret, node, popped));
break;
case NODE_WHILE:
case NODE_UNTIL:
CHECK(compile_loop(iseq, ret, node, popped, type));
break;
case NODE_FOR:
case NODE_ITER:
CHECK(compile_iter(iseq, ret, node, popped));
break;
case NODE_FOR_MASGN:
CHECK(compile_for_masgn(iseq, ret, node, popped));
break;
case NODE_BREAK:
CHECK(compile_break(iseq, ret, node, popped));
break;
case NODE_NEXT:
CHECK(compile_next(iseq, ret, node, popped));
break;
case NODE_REDO:
CHECK(compile_redo(iseq, ret, node, popped));
break;
case NODE_RETRY:
CHECK(compile_retry(iseq, ret, node, popped));
break;
case NODE_BEGIN:{
CHECK(COMPILE_(ret, "NODE_BEGIN", node->nd_body, popped));
break;
}
case NODE_RESCUE:
CHECK(compile_rescue(iseq, ret, node, popped));
break;
case NODE_RESBODY:
CHECK(compile_resbody(iseq, ret, node, popped));
break;
case NODE_ENSURE:
CHECK(compile_ensure(iseq, ret, node, popped));
break;
case NODE_AND:
case NODE_OR:{
LABEL *end_label = NEW_LABEL(line);
CHECK(COMPILE(ret, "nd_1st", node->nd_1st));
if (!popped) {
ADD_INSN(ret, line, dup);
}
if (type == NODE_AND) {
ADD_INSNL(ret, line, branchunless, end_label);
}
else {
ADD_INSNL(ret, line, branchif, end_label);
}
if (!popped) {
ADD_INSN(ret, line, pop);
}
CHECK(COMPILE_(ret, "nd_2nd", node->nd_2nd, popped));
ADD_LABEL(ret, end_label);
break;
}
case NODE_MASGN:{
compile_massign(iseq, ret, node, popped);
break;
}
case NODE_LASGN:{
ID id = node->nd_vid;
int idx = body->local_iseq->body->local_table_size - get_local_var_idx(iseq, id);
debugs("lvar: %s idx: %d\n", rb_id2name(id), idx);
CHECK(COMPILE(ret, "rvalue", node->nd_value));
if (!popped) {
ADD_INSN(ret, line, dup);
}
ADD_SETLOCAL(ret, line, idx, get_lvar_level(iseq));
break;
}
case NODE_DASGN:
case NODE_DASGN_CURR:{
int idx, lv, ls;
ID id = node->nd_vid;
CHECK(COMPILE(ret, "dvalue", node->nd_value));
debugi("dassn id", rb_id2str(id) ? id : '*');
if (!popped) {
ADD_INSN(ret, line, dup);
}
idx = get_dyna_var_idx(iseq, id, &lv, &ls);
if (idx < 0) {
COMPILE_ERROR(ERROR_ARGS "NODE_DASGN(_CURR): unknown id (%"PRIsVALUE")",
rb_id2str(id));
goto ng;
}
ADD_SETLOCAL(ret, line, ls - idx, lv);
break;
}
case NODE_GASGN:{
CHECK(COMPILE(ret, "lvalue", node->nd_value));
if (!popped) {
ADD_INSN(ret, line, dup);
}
ADD_INSN1(ret, line, setglobal,
((VALUE)node->nd_entry | 1));
break;
}
case NODE_IASGN:{
CHECK(COMPILE(ret, "lvalue", node->nd_value));
if (!popped) {
ADD_INSN(ret, line, dup);
}
ADD_INSN2(ret, line, setinstancevariable,
ID2SYM(node->nd_vid),
get_ivar_ic_value(iseq,node->nd_vid));
break;
}
case NODE_CDECL:{
2019-06-16 17:01:46 +03:00
CHECK(COMPILE(ret, "lvalue", node->nd_value));
2019-06-16 17:01:46 +03:00
if (!popped) {
ADD_INSN(ret, line, dup);
}
if (node->nd_vid) {
ADD_INSN1(ret, line, putspecialobject,
INT2FIX(VM_SPECIAL_OBJECT_CONST_BASE));
2019-06-16 17:01:46 +03:00
ADD_INSN1(ret, line, setconstant, ID2SYM(node->nd_vid));
}
else {
compile_cpath(ret, iseq, node->nd_else);
2019-06-16 17:01:46 +03:00
ADD_INSN1(ret, line, setconstant, ID2SYM(node->nd_else->nd_mid));
}
break;
}
case NODE_CVASGN:{
CHECK(COMPILE(ret, "cvasgn val", node->nd_value));
if (!popped) {
ADD_INSN(ret, line, dup);
}
ADD_INSN1(ret, line, setclassvariable,
ID2SYM(node->nd_vid));
break;
}
case NODE_OP_ASGN1: {
VALUE argc;
unsigned int flag = 0;
2019-09-19 17:44:37 +03:00
int asgnflag = 0;
ID id = node->nd_mid;
int boff = 0;
/*
* a[x] (op)= y
*
* nil # nil
* eval a # nil a
* eval x # nil a x
* dupn 2 # nil a x a x
* send :[] # nil a x a[x]
* eval y # nil a x a[x] y
* send op # nil a x ret
* setn 3 # ret a x ret
* send []= # ret ?
* pop # ret
*/
/*
* nd_recv[nd_args->nd_body] (nd_mid)= nd_args->nd_head;
* NODE_OP_ASGN nd_recv
* nd_args->nd_head
* nd_args->nd_body
* nd_mid
*/
if (!popped) {
ADD_INSN(ret, line, putnil);
}
asgnflag = COMPILE_RECV(ret, "NODE_OP_ASGN1 recv", node);
2019-09-19 17:44:37 +03:00
CHECK(asgnflag != -1);
switch (nd_type(node->nd_args->nd_head)) {
case NODE_ZLIST:
argc = INT2FIX(0);
break;
case NODE_BLOCK_PASS:
boff = 1;
2019-07-14 18:10:39 +03:00
/* fall through */
default:
argc = setup_args(iseq, ret, node->nd_args->nd_head, &flag, NULL);
CHECK(!NIL_P(argc));
}
ADD_INSN1(ret, line, dupn, FIXNUM_INC(argc, 1 + boff));
flag |= asgnflag;
* rewrite method/block parameter fitting logic to optimize keyword arguments/parameters and a splat argument. [Feature #10440] (Details are described in this ticket) Most of complex part is moved to vm_args.c. Now, ISeq#to_a does not catch up new instruction format. * vm_core.h: change iseq data structures. * introduce rb_call_info_kw_arg_t to represent keyword arguments. * add rb_call_info_t::kw_arg. * rename rb_iseq_t::arg_post_len to rb_iseq_t::arg_post_num. * rename rb_iseq_t::arg_keywords to arg_keyword_num. * rename rb_iseq_t::arg_keyword to rb_iseq_t::arg_keyword_bits. to represent keyword bitmap parameter index. This bitmap parameter shows that which keyword parameters are given or not given (0 for given). It is refered by `checkkeyword' instruction described bellow. * rename rb_iseq_t::arg_keyword_check to rb_iseq_t::arg_keyword_rest to represent keyword rest parameter index. * add rb_iseq_t::arg_keyword_default_values to represent default keyword values. * rename VM_CALL_ARGS_SKIP_SETUP to VM_CALL_ARGS_SIMPLE to represent (ci->flag & (SPLAT|BLOCKARG)) && ci->blockiseq == NULL && ci->kw_arg == NULL. * vm_insnhelper.c, vm_args.c: rewrite with refactoring. * rewrite splat argument code. * rewrite keyword arguments/parameters code. * merge method and block parameter fitting code into one code base. * vm.c, vm_eval.c: catch up these changes. * compile.c (new_callinfo): callinfo requires kw_arg parameter. * compile.c (compile_array_): check the last argument Hash object or not. If Hash object and all keys are Symbol literals, they are compiled to keyword arguments. * insns.def (checkkeyword): add new instruction. This instruction check the availability of corresponding keyword. For example, a method "def foo k1: 'v1'; end" is cimpiled to the following instructions. 0000 checkkeyword 2, 0 # check k1 is given. 0003 branchif 9 # if given, jump to address #9 0005 putstring "v1" 0007 setlocal_OP__WC__0 3 # k1 = 'v1' 0009 trace 8 0011 putnil 0012 trace 16 0014 leave * insns.def (opt_send_simple): removed and add new instruction "opt_send_without_block". * parse.y (new_args_tail_gen): reorder variables. Before this patch, a method "def foo(k1: 1, kr1:, k2: 2, **krest, &b)" has parameter variables "k1, kr1, k2, &b, internal_id, krest", but this patch reorders to "kr1, k1, k2, internal_id, krest, &b". (locate a block variable at last) * parse.y (vtable_pop): added. This function remove latest `n' variables from vtable. * iseq.c: catch up iseq data changes. * proc.c: ditto. * class.c (keyword_error): export as rb_keyword_error(). * common.mk: depend vm_args.c for vm.o. * hash.c (rb_hash_has_key): export. * internal.h: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@48239 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2014-11-02 21:02:55 +03:00
ADD_SEND_WITH_FLAG(ret, line, idAREF, argc, INT2FIX(flag));
if (id == idOROP || id == idANDOP) {
/* a[x] ||= y or a[x] &&= y
unless/if a[x]
a[x]= y
else
nil
end
*/
LABEL *label = NEW_LABEL(line);
LABEL *lfin = NEW_LABEL(line);
ADD_INSN(ret, line, dup);
if (id == idOROP) {
ADD_INSNL(ret, line, branchif, label);
}
else { /* idANDOP */
ADD_INSNL(ret, line, branchunless, label);
}
ADD_INSN(ret, line, pop);
CHECK(COMPILE(ret, "NODE_OP_ASGN1 args->body: ", node->nd_args->nd_body));
if (!popped) {
ADD_INSN1(ret, line, setn, FIXNUM_INC(argc, 2+boff));
}
if (flag & VM_CALL_ARGS_SPLAT) {
ADD_INSN1(ret, line, newarray, INT2FIX(1));
if (boff > 0) {
ADD_INSN1(ret, line, dupn, INT2FIX(3));
ADD_INSN(ret, line, swap);
ADD_INSN(ret, line, pop);
}
ADD_INSN(ret, line, concatarray);
if (boff > 0) {
ADD_INSN1(ret, line, setn, INT2FIX(3));
ADD_INSN(ret, line, pop);
ADD_INSN(ret, line, pop);
}
* rewrite method/block parameter fitting logic to optimize keyword arguments/parameters and a splat argument. [Feature #10440] (Details are described in this ticket) Most of complex part is moved to vm_args.c. Now, ISeq#to_a does not catch up new instruction format. * vm_core.h: change iseq data structures. * introduce rb_call_info_kw_arg_t to represent keyword arguments. * add rb_call_info_t::kw_arg. * rename rb_iseq_t::arg_post_len to rb_iseq_t::arg_post_num. * rename rb_iseq_t::arg_keywords to arg_keyword_num. * rename rb_iseq_t::arg_keyword to rb_iseq_t::arg_keyword_bits. to represent keyword bitmap parameter index. This bitmap parameter shows that which keyword parameters are given or not given (0 for given). It is refered by `checkkeyword' instruction described bellow. * rename rb_iseq_t::arg_keyword_check to rb_iseq_t::arg_keyword_rest to represent keyword rest parameter index. * add rb_iseq_t::arg_keyword_default_values to represent default keyword values. * rename VM_CALL_ARGS_SKIP_SETUP to VM_CALL_ARGS_SIMPLE to represent (ci->flag & (SPLAT|BLOCKARG)) && ci->blockiseq == NULL && ci->kw_arg == NULL. * vm_insnhelper.c, vm_args.c: rewrite with refactoring. * rewrite splat argument code. * rewrite keyword arguments/parameters code. * merge method and block parameter fitting code into one code base. * vm.c, vm_eval.c: catch up these changes. * compile.c (new_callinfo): callinfo requires kw_arg parameter. * compile.c (compile_array_): check the last argument Hash object or not. If Hash object and all keys are Symbol literals, they are compiled to keyword arguments. * insns.def (checkkeyword): add new instruction. This instruction check the availability of corresponding keyword. For example, a method "def foo k1: 'v1'; end" is cimpiled to the following instructions. 0000 checkkeyword 2, 0 # check k1 is given. 0003 branchif 9 # if given, jump to address #9 0005 putstring "v1" 0007 setlocal_OP__WC__0 3 # k1 = 'v1' 0009 trace 8 0011 putnil 0012 trace 16 0014 leave * insns.def (opt_send_simple): removed and add new instruction "opt_send_without_block". * parse.y (new_args_tail_gen): reorder variables. Before this patch, a method "def foo(k1: 1, kr1:, k2: 2, **krest, &b)" has parameter variables "k1, kr1, k2, &b, internal_id, krest", but this patch reorders to "kr1, k1, k2, internal_id, krest, &b". (locate a block variable at last) * parse.y (vtable_pop): added. This function remove latest `n' variables from vtable. * iseq.c: catch up iseq data changes. * proc.c: ditto. * class.c (keyword_error): export as rb_keyword_error(). * common.mk: depend vm_args.c for vm.o. * hash.c (rb_hash_has_key): export. * internal.h: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@48239 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2014-11-02 21:02:55 +03:00
ADD_SEND_WITH_FLAG(ret, line, idASET, argc, INT2FIX(flag));
}
else {
if (boff > 0)
ADD_INSN(ret, line, swap);
* rewrite method/block parameter fitting logic to optimize keyword arguments/parameters and a splat argument. [Feature #10440] (Details are described in this ticket) Most of complex part is moved to vm_args.c. Now, ISeq#to_a does not catch up new instruction format. * vm_core.h: change iseq data structures. * introduce rb_call_info_kw_arg_t to represent keyword arguments. * add rb_call_info_t::kw_arg. * rename rb_iseq_t::arg_post_len to rb_iseq_t::arg_post_num. * rename rb_iseq_t::arg_keywords to arg_keyword_num. * rename rb_iseq_t::arg_keyword to rb_iseq_t::arg_keyword_bits. to represent keyword bitmap parameter index. This bitmap parameter shows that which keyword parameters are given or not given (0 for given). It is refered by `checkkeyword' instruction described bellow. * rename rb_iseq_t::arg_keyword_check to rb_iseq_t::arg_keyword_rest to represent keyword rest parameter index. * add rb_iseq_t::arg_keyword_default_values to represent default keyword values. * rename VM_CALL_ARGS_SKIP_SETUP to VM_CALL_ARGS_SIMPLE to represent (ci->flag & (SPLAT|BLOCKARG)) && ci->blockiseq == NULL && ci->kw_arg == NULL. * vm_insnhelper.c, vm_args.c: rewrite with refactoring. * rewrite splat argument code. * rewrite keyword arguments/parameters code. * merge method and block parameter fitting code into one code base. * vm.c, vm_eval.c: catch up these changes. * compile.c (new_callinfo): callinfo requires kw_arg parameter. * compile.c (compile_array_): check the last argument Hash object or not. If Hash object and all keys are Symbol literals, they are compiled to keyword arguments. * insns.def (checkkeyword): add new instruction. This instruction check the availability of corresponding keyword. For example, a method "def foo k1: 'v1'; end" is cimpiled to the following instructions. 0000 checkkeyword 2, 0 # check k1 is given. 0003 branchif 9 # if given, jump to address #9 0005 putstring "v1" 0007 setlocal_OP__WC__0 3 # k1 = 'v1' 0009 trace 8 0011 putnil 0012 trace 16 0014 leave * insns.def (opt_send_simple): removed and add new instruction "opt_send_without_block". * parse.y (new_args_tail_gen): reorder variables. Before this patch, a method "def foo(k1: 1, kr1:, k2: 2, **krest, &b)" has parameter variables "k1, kr1, k2, &b, internal_id, krest", but this patch reorders to "kr1, k1, k2, internal_id, krest, &b". (locate a block variable at last) * parse.y (vtable_pop): added. This function remove latest `n' variables from vtable. * iseq.c: catch up iseq data changes. * proc.c: ditto. * class.c (keyword_error): export as rb_keyword_error(). * common.mk: depend vm_args.c for vm.o. * hash.c (rb_hash_has_key): export. * internal.h: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@48239 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2014-11-02 21:02:55 +03:00
ADD_SEND_WITH_FLAG(ret, line, idASET, FIXNUM_INC(argc, 1), INT2FIX(flag));
}
ADD_INSN(ret, line, pop);
ADD_INSNL(ret, line, jump, lfin);
ADD_LABEL(ret, label);
if (!popped) {
ADD_INSN1(ret, line, setn, FIXNUM_INC(argc, 2+boff));
}
ADD_INSN1(ret, line, adjuststack, FIXNUM_INC(argc, 2+boff));
ADD_LABEL(ret, lfin);
}
else {
CHECK(COMPILE(ret, "NODE_OP_ASGN1 args->body: ", node->nd_args->nd_body));
ADD_SEND(ret, line, id, INT2FIX(1));
if (!popped) {
ADD_INSN1(ret, line, setn, FIXNUM_INC(argc, 2+boff));
}
if (flag & VM_CALL_ARGS_SPLAT) {
ADD_INSN1(ret, line, newarray, INT2FIX(1));
if (boff > 0) {
ADD_INSN1(ret, line, dupn, INT2FIX(3));
ADD_INSN(ret, line, swap);
ADD_INSN(ret, line, pop);
}
ADD_INSN(ret, line, concatarray);
if (boff > 0) {
ADD_INSN1(ret, line, setn, INT2FIX(3));
ADD_INSN(ret, line, pop);
ADD_INSN(ret, line, pop);
}
* rewrite method/block parameter fitting logic to optimize keyword arguments/parameters and a splat argument. [Feature #10440] (Details are described in this ticket) Most of complex part is moved to vm_args.c. Now, ISeq#to_a does not catch up new instruction format. * vm_core.h: change iseq data structures. * introduce rb_call_info_kw_arg_t to represent keyword arguments. * add rb_call_info_t::kw_arg. * rename rb_iseq_t::arg_post_len to rb_iseq_t::arg_post_num. * rename rb_iseq_t::arg_keywords to arg_keyword_num. * rename rb_iseq_t::arg_keyword to rb_iseq_t::arg_keyword_bits. to represent keyword bitmap parameter index. This bitmap parameter shows that which keyword parameters are given or not given (0 for given). It is refered by `checkkeyword' instruction described bellow. * rename rb_iseq_t::arg_keyword_check to rb_iseq_t::arg_keyword_rest to represent keyword rest parameter index. * add rb_iseq_t::arg_keyword_default_values to represent default keyword values. * rename VM_CALL_ARGS_SKIP_SETUP to VM_CALL_ARGS_SIMPLE to represent (ci->flag & (SPLAT|BLOCKARG)) && ci->blockiseq == NULL && ci->kw_arg == NULL. * vm_insnhelper.c, vm_args.c: rewrite with refactoring. * rewrite splat argument code. * rewrite keyword arguments/parameters code. * merge method and block parameter fitting code into one code base. * vm.c, vm_eval.c: catch up these changes. * compile.c (new_callinfo): callinfo requires kw_arg parameter. * compile.c (compile_array_): check the last argument Hash object or not. If Hash object and all keys are Symbol literals, they are compiled to keyword arguments. * insns.def (checkkeyword): add new instruction. This instruction check the availability of corresponding keyword. For example, a method "def foo k1: 'v1'; end" is cimpiled to the following instructions. 0000 checkkeyword 2, 0 # check k1 is given. 0003 branchif 9 # if given, jump to address #9 0005 putstring "v1" 0007 setlocal_OP__WC__0 3 # k1 = 'v1' 0009 trace 8 0011 putnil 0012 trace 16 0014 leave * insns.def (opt_send_simple): removed and add new instruction "opt_send_without_block". * parse.y (new_args_tail_gen): reorder variables. Before this patch, a method "def foo(k1: 1, kr1:, k2: 2, **krest, &b)" has parameter variables "k1, kr1, k2, &b, internal_id, krest", but this patch reorders to "kr1, k1, k2, internal_id, krest, &b". (locate a block variable at last) * parse.y (vtable_pop): added. This function remove latest `n' variables from vtable. * iseq.c: catch up iseq data changes. * proc.c: ditto. * class.c (keyword_error): export as rb_keyword_error(). * common.mk: depend vm_args.c for vm.o. * hash.c (rb_hash_has_key): export. * internal.h: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@48239 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2014-11-02 21:02:55 +03:00
ADD_SEND_WITH_FLAG(ret, line, idASET, argc, INT2FIX(flag));
}
else {
if (boff > 0)
ADD_INSN(ret, line, swap);
* rewrite method/block parameter fitting logic to optimize keyword arguments/parameters and a splat argument. [Feature #10440] (Details are described in this ticket) Most of complex part is moved to vm_args.c. Now, ISeq#to_a does not catch up new instruction format. * vm_core.h: change iseq data structures. * introduce rb_call_info_kw_arg_t to represent keyword arguments. * add rb_call_info_t::kw_arg. * rename rb_iseq_t::arg_post_len to rb_iseq_t::arg_post_num. * rename rb_iseq_t::arg_keywords to arg_keyword_num. * rename rb_iseq_t::arg_keyword to rb_iseq_t::arg_keyword_bits. to represent keyword bitmap parameter index. This bitmap parameter shows that which keyword parameters are given or not given (0 for given). It is refered by `checkkeyword' instruction described bellow. * rename rb_iseq_t::arg_keyword_check to rb_iseq_t::arg_keyword_rest to represent keyword rest parameter index. * add rb_iseq_t::arg_keyword_default_values to represent default keyword values. * rename VM_CALL_ARGS_SKIP_SETUP to VM_CALL_ARGS_SIMPLE to represent (ci->flag & (SPLAT|BLOCKARG)) && ci->blockiseq == NULL && ci->kw_arg == NULL. * vm_insnhelper.c, vm_args.c: rewrite with refactoring. * rewrite splat argument code. * rewrite keyword arguments/parameters code. * merge method and block parameter fitting code into one code base. * vm.c, vm_eval.c: catch up these changes. * compile.c (new_callinfo): callinfo requires kw_arg parameter. * compile.c (compile_array_): check the last argument Hash object or not. If Hash object and all keys are Symbol literals, they are compiled to keyword arguments. * insns.def (checkkeyword): add new instruction. This instruction check the availability of corresponding keyword. For example, a method "def foo k1: 'v1'; end" is cimpiled to the following instructions. 0000 checkkeyword 2, 0 # check k1 is given. 0003 branchif 9 # if given, jump to address #9 0005 putstring "v1" 0007 setlocal_OP__WC__0 3 # k1 = 'v1' 0009 trace 8 0011 putnil 0012 trace 16 0014 leave * insns.def (opt_send_simple): removed and add new instruction "opt_send_without_block". * parse.y (new_args_tail_gen): reorder variables. Before this patch, a method "def foo(k1: 1, kr1:, k2: 2, **krest, &b)" has parameter variables "k1, kr1, k2, &b, internal_id, krest", but this patch reorders to "kr1, k1, k2, internal_id, krest, &b". (locate a block variable at last) * parse.y (vtable_pop): added. This function remove latest `n' variables from vtable. * iseq.c: catch up iseq data changes. * proc.c: ditto. * class.c (keyword_error): export as rb_keyword_error(). * common.mk: depend vm_args.c for vm.o. * hash.c (rb_hash_has_key): export. * internal.h: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@48239 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2014-11-02 21:02:55 +03:00
ADD_SEND_WITH_FLAG(ret, line, idASET, FIXNUM_INC(argc, 1), INT2FIX(flag));
}
ADD_INSN(ret, line, pop);
}
break;
}
case NODE_OP_ASGN2:{
ID atype = node->nd_next->nd_mid;
ID vid = node->nd_next->nd_vid, aid = rb_id_attrset(vid);
2019-09-19 17:44:37 +03:00
int asgnflag;
LABEL *lfin = NEW_LABEL(line);
LABEL *lcfin = NEW_LABEL(line);
LABEL *lskip = 0;
/*
class C; attr_accessor :c; end
r = C.new
r.a &&= v # asgn2
eval r # r
dup # r r
eval r.a # r o
# or
dup # r o o
if lcfin # r o
pop # r
eval v # r v
swap # v r
topn 1 # v r v
send a= # v ?
jump lfin # v ?
lcfin: # r o
swap # o r
lfin: # o ?
pop # o
# and
dup # r o o
unless lcfin
pop # r
eval v # r v
swap # v r
topn 1 # v r v
send a= # v ?
jump lfin # v ?
# others
eval v # r o v
send ?? # r w
send a= # w
*/
asgnflag = COMPILE_RECV(ret, "NODE_OP_ASGN2#recv", node);
2019-09-19 17:44:37 +03:00
CHECK(asgnflag != -1);
if (node->nd_next->nd_aid) {
lskip = NEW_LABEL(line);
ADD_INSN(ret, line, dup);
ADD_INSNL(ret, line, branchnil, lskip);
}
ADD_INSN(ret, line, dup);
ADD_SEND_WITH_FLAG(ret, line, vid, INT2FIX(0), INT2FIX(asgnflag));
if (atype == idOROP || atype == idANDOP) {
ADD_INSN(ret, line, dup);
if (atype == idOROP) {
ADD_INSNL(ret, line, branchif, lcfin);
}
else { /* idANDOP */
ADD_INSNL(ret, line, branchunless, lcfin);
}
ADD_INSN(ret, line, pop);
CHECK(COMPILE(ret, "NODE_OP_ASGN2 val", node->nd_value));
ADD_INSN(ret, line, swap);
ADD_INSN1(ret, line, topn, INT2FIX(1));
ADD_SEND_WITH_FLAG(ret, line, aid, INT2FIX(1), INT2FIX(asgnflag));
ADD_INSNL(ret, line, jump, lfin);
ADD_LABEL(ret, lcfin);
ADD_INSN(ret, line, swap);
ADD_LABEL(ret, lfin);
ADD_INSN(ret, line, pop);
if (lskip) {
ADD_LABEL(ret, lskip);
}
if (popped) {
/* we can apply more optimize */
ADD_INSN(ret, line, pop);
}
}
else {
CHECK(COMPILE(ret, "NODE_OP_ASGN2 val", node->nd_value));
ADD_SEND(ret, line, atype, INT2FIX(1));
if (!popped) {
ADD_INSN(ret, line, swap);
ADD_INSN1(ret, line, topn, INT2FIX(1));
}
ADD_SEND_WITH_FLAG(ret, line, aid, INT2FIX(1), INT2FIX(asgnflag));
if (lskip && popped) {
ADD_LABEL(ret, lskip);
}
ADD_INSN(ret, line, pop);
if (lskip && !popped) {
ADD_LABEL(ret, lskip);
}
}
break;
}
case NODE_OP_CDECL: {
LABEL *lfin = 0;
LABEL *lassign = 0;
ID mid;
switch (nd_type(node->nd_head)) {
case NODE_COLON3:
ADD_INSN1(ret, line, putobject, rb_cObject);
break;
case NODE_COLON2:
CHECK(COMPILE(ret, "NODE_OP_CDECL/colon2#nd_head", node->nd_head->nd_head));
break;
default:
COMPILE_ERROR(ERROR_ARGS "%s: invalid node in NODE_OP_CDECL",
ruby_node_name(nd_type(node->nd_head)));
goto ng;
}
mid = node->nd_head->nd_mid;
/* cref */
if (node->nd_aid == idOROP) {
lassign = NEW_LABEL(line);
ADD_INSN(ret, line, dup); /* cref cref */
ADD_INSN3(ret, line, defined, INT2FIX(DEFINED_CONST),
ID2SYM(mid), Qfalse); /* cref bool */
ADD_INSNL(ret, line, branchunless, lassign); /* cref */
}
ADD_INSN(ret, line, dup); /* cref cref */
ADD_INSN1(ret, line, putobject, Qtrue);
ADD_INSN1(ret, line, getconstant, ID2SYM(mid)); /* cref obj */
if (node->nd_aid == idOROP || node->nd_aid == idANDOP) {
lfin = NEW_LABEL(line);
if (!popped) ADD_INSN(ret, line, dup); /* cref [obj] obj */
if (node->nd_aid == idOROP)
ADD_INSNL(ret, line, branchif, lfin);
else /* idANDOP */
ADD_INSNL(ret, line, branchunless, lfin);
/* cref [obj] */
if (!popped) ADD_INSN(ret, line, pop); /* cref */
if (lassign) ADD_LABEL(ret, lassign);
CHECK(COMPILE(ret, "NODE_OP_CDECL#nd_value", node->nd_value));
/* cref value */
if (popped)
ADD_INSN1(ret, line, topn, INT2FIX(1)); /* cref value cref */
else {
ADD_INSN1(ret, line, dupn, INT2FIX(2)); /* cref value cref value */
ADD_INSN(ret, line, swap); /* cref value value cref */
}
ADD_INSN1(ret, line, setconstant, ID2SYM(mid)); /* cref [value] */
ADD_LABEL(ret, lfin); /* cref [value] */
if (!popped) ADD_INSN(ret, line, swap); /* [value] cref */
ADD_INSN(ret, line, pop); /* [value] */
}
else {
CHECK(COMPILE(ret, "NODE_OP_CDECL#nd_value", node->nd_value));
/* cref obj value */
ADD_CALL(ret, line, node->nd_aid, INT2FIX(1));
/* cref value */
ADD_INSN(ret, line, swap); /* value cref */
if (!popped) {
ADD_INSN1(ret, line, topn, INT2FIX(1)); /* value cref value */
ADD_INSN(ret, line, swap); /* value value cref */
}
ADD_INSN1(ret, line, setconstant, ID2SYM(mid));
}
break;
}
case NODE_OP_ASGN_AND:
case NODE_OP_ASGN_OR:{
LABEL *lfin = NEW_LABEL(line);
LABEL *lassign;
if (nd_type(node) == NODE_OP_ASGN_OR) {
LABEL *lfinish[2];
lfinish[0] = lfin;
lfinish[1] = 0;
defined_expr(iseq, ret, node->nd_head, lfinish, Qfalse);
lassign = lfinish[1];
if (!lassign) {
lassign = NEW_LABEL(line);
}
ADD_INSNL(ret, line, branchunless, lassign);
}
else {
lassign = NEW_LABEL(line);
}
CHECK(COMPILE(ret, "NODE_OP_ASGN_AND/OR#nd_head", node->nd_head));
ADD_INSN(ret, line, dup);
if (nd_type(node) == NODE_OP_ASGN_AND) {
ADD_INSNL(ret, line, branchunless, lfin);
}
else {
ADD_INSNL(ret, line, branchif, lfin);
}
ADD_INSN(ret, line, pop);
ADD_LABEL(ret, lassign);
CHECK(COMPILE(ret, "NODE_OP_ASGN_AND/OR#nd_value", node->nd_value));
ADD_LABEL(ret, lfin);
if (popped) {
/* we can apply more optimize */
ADD_INSN(ret, line, pop);
}
break;
}
case NODE_CALL: /* obj.foo */
case NODE_OPCALL: /* foo[] */
if (compile_call_precheck_freeze(iseq, ret, node, line, popped) == TRUE) {
break;
}
case NODE_QCALL: /* obj&.foo */
case NODE_FCALL: /* foo() */
case NODE_VCALL: /* foo (variable or call) */
if (compile_call(iseq, ret, node, type, line, popped) == COMPILE_NG) {
goto ng;
}
break;
case NODE_SUPER:
case NODE_ZSUPER:{
DECL_ANCHOR(args);
int argc;
unsigned int flag = 0;
struct rb_call_info_kw_arg *keywords = NULL;
const rb_iseq_t *parent_block = ISEQ_COMPILE_DATA(iseq)->current_block;
INIT_ANCHOR(args);
ISEQ_COMPILE_DATA(iseq)->current_block = NULL;
if (type == NODE_SUPER) {
* rewrite method/block parameter fitting logic to optimize keyword arguments/parameters and a splat argument. [Feature #10440] (Details are described in this ticket) Most of complex part is moved to vm_args.c. Now, ISeq#to_a does not catch up new instruction format. * vm_core.h: change iseq data structures. * introduce rb_call_info_kw_arg_t to represent keyword arguments. * add rb_call_info_t::kw_arg. * rename rb_iseq_t::arg_post_len to rb_iseq_t::arg_post_num. * rename rb_iseq_t::arg_keywords to arg_keyword_num. * rename rb_iseq_t::arg_keyword to rb_iseq_t::arg_keyword_bits. to represent keyword bitmap parameter index. This bitmap parameter shows that which keyword parameters are given or not given (0 for given). It is refered by `checkkeyword' instruction described bellow. * rename rb_iseq_t::arg_keyword_check to rb_iseq_t::arg_keyword_rest to represent keyword rest parameter index. * add rb_iseq_t::arg_keyword_default_values to represent default keyword values. * rename VM_CALL_ARGS_SKIP_SETUP to VM_CALL_ARGS_SIMPLE to represent (ci->flag & (SPLAT|BLOCKARG)) && ci->blockiseq == NULL && ci->kw_arg == NULL. * vm_insnhelper.c, vm_args.c: rewrite with refactoring. * rewrite splat argument code. * rewrite keyword arguments/parameters code. * merge method and block parameter fitting code into one code base. * vm.c, vm_eval.c: catch up these changes. * compile.c (new_callinfo): callinfo requires kw_arg parameter. * compile.c (compile_array_): check the last argument Hash object or not. If Hash object and all keys are Symbol literals, they are compiled to keyword arguments. * insns.def (checkkeyword): add new instruction. This instruction check the availability of corresponding keyword. For example, a method "def foo k1: 'v1'; end" is cimpiled to the following instructions. 0000 checkkeyword 2, 0 # check k1 is given. 0003 branchif 9 # if given, jump to address #9 0005 putstring "v1" 0007 setlocal_OP__WC__0 3 # k1 = 'v1' 0009 trace 8 0011 putnil 0012 trace 16 0014 leave * insns.def (opt_send_simple): removed and add new instruction "opt_send_without_block". * parse.y (new_args_tail_gen): reorder variables. Before this patch, a method "def foo(k1: 1, kr1:, k2: 2, **krest, &b)" has parameter variables "k1, kr1, k2, &b, internal_id, krest", but this patch reorders to "kr1, k1, k2, internal_id, krest, &b". (locate a block variable at last) * parse.y (vtable_pop): added. This function remove latest `n' variables from vtable. * iseq.c: catch up iseq data changes. * proc.c: ditto. * class.c (keyword_error): export as rb_keyword_error(). * common.mk: depend vm_args.c for vm.o. * hash.c (rb_hash_has_key): export. * internal.h: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@48239 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2014-11-02 21:02:55 +03:00
VALUE vargc = setup_args(iseq, args, node->nd_args, &flag, &keywords);
CHECK(!NIL_P(vargc));
argc = FIX2INT(vargc);
}
else {
/* NODE_ZSUPER */
int i;
const rb_iseq_t *liseq = body->local_iseq;
const struct rb_iseq_constant_body *const local_body = liseq->body;
const struct rb_iseq_param_keyword *const local_kwd = local_body->param.keyword;
int lvar_level = get_lvar_level(iseq);
argc = local_body->param.lead_num;
/* normal arguments */
for (i = 0; i < local_body->param.lead_num; i++) {
int idx = local_body->local_table_size - i;
ADD_GETLOCAL(args, line, idx, lvar_level);
}
if (local_body->param.flags.has_opt) {
/* optional arguments */
int j;
for (j = 0; j < local_body->param.opt_num; j++) {
int idx = local_body->local_table_size - (i + j);
ADD_GETLOCAL(args, line, idx, lvar_level);
}
i += j;
argc = i;
}
if (local_body->param.flags.has_rest) {
/* rest argument */
int idx = local_body->local_table_size - local_body->param.rest_start;
ADD_GETLOCAL(args, line, idx, lvar_level);
ADD_INSN1(args, line, splatarray, Qfalse);
argc = local_body->param.rest_start + 1;
flag |= VM_CALL_ARGS_SPLAT;
}
if (local_body->param.flags.has_post) {
/* post arguments */
int post_len = local_body->param.post_num;
int post_start = local_body->param.post_start;
if (local_body->param.flags.has_rest) {
int j;
for (j=0; j<post_len; j++) {
int idx = local_body->local_table_size - (post_start + j);
ADD_GETLOCAL(args, line, idx, lvar_level);
}
ADD_INSN1(args, line, newarray, INT2FIX(j));
ADD_INSN (args, line, concatarray);
/* argc is settled at above */
}
else {
int j;
for (j=0; j<post_len; j++) {
int idx = local_body->local_table_size - (post_start + j);
ADD_GETLOCAL(args, line, idx, lvar_level);
}
argc = post_len + post_start;
}
}
if (local_body->param.flags.has_kw) { /* TODO: support keywords */
int local_size = local_body->local_table_size;
argc++;
* rewrite method/block parameter fitting logic to optimize keyword arguments/parameters and a splat argument. [Feature #10440] (Details are described in this ticket) Most of complex part is moved to vm_args.c. Now, ISeq#to_a does not catch up new instruction format. * vm_core.h: change iseq data structures. * introduce rb_call_info_kw_arg_t to represent keyword arguments. * add rb_call_info_t::kw_arg. * rename rb_iseq_t::arg_post_len to rb_iseq_t::arg_post_num. * rename rb_iseq_t::arg_keywords to arg_keyword_num. * rename rb_iseq_t::arg_keyword to rb_iseq_t::arg_keyword_bits. to represent keyword bitmap parameter index. This bitmap parameter shows that which keyword parameters are given or not given (0 for given). It is refered by `checkkeyword' instruction described bellow. * rename rb_iseq_t::arg_keyword_check to rb_iseq_t::arg_keyword_rest to represent keyword rest parameter index. * add rb_iseq_t::arg_keyword_default_values to represent default keyword values. * rename VM_CALL_ARGS_SKIP_SETUP to VM_CALL_ARGS_SIMPLE to represent (ci->flag & (SPLAT|BLOCKARG)) && ci->blockiseq == NULL && ci->kw_arg == NULL. * vm_insnhelper.c, vm_args.c: rewrite with refactoring. * rewrite splat argument code. * rewrite keyword arguments/parameters code. * merge method and block parameter fitting code into one code base. * vm.c, vm_eval.c: catch up these changes. * compile.c (new_callinfo): callinfo requires kw_arg parameter. * compile.c (compile_array_): check the last argument Hash object or not. If Hash object and all keys are Symbol literals, they are compiled to keyword arguments. * insns.def (checkkeyword): add new instruction. This instruction check the availability of corresponding keyword. For example, a method "def foo k1: 'v1'; end" is cimpiled to the following instructions. 0000 checkkeyword 2, 0 # check k1 is given. 0003 branchif 9 # if given, jump to address #9 0005 putstring "v1" 0007 setlocal_OP__WC__0 3 # k1 = 'v1' 0009 trace 8 0011 putnil 0012 trace 16 0014 leave * insns.def (opt_send_simple): removed and add new instruction "opt_send_without_block". * parse.y (new_args_tail_gen): reorder variables. Before this patch, a method "def foo(k1: 1, kr1:, k2: 2, **krest, &b)" has parameter variables "k1, kr1, k2, &b, internal_id, krest", but this patch reorders to "kr1, k1, k2, internal_id, krest, &b". (locate a block variable at last) * parse.y (vtable_pop): added. This function remove latest `n' variables from vtable. * iseq.c: catch up iseq data changes. * proc.c: ditto. * class.c (keyword_error): export as rb_keyword_error(). * common.mk: depend vm_args.c for vm.o. * hash.c (rb_hash_has_key): export. * internal.h: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@48239 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2014-11-02 21:02:55 +03:00
ADD_INSN1(args, line, putspecialobject, INT2FIX(VM_SPECIAL_OBJECT_VMCORE));
* rewrite method/block parameter fitting logic to optimize keyword arguments/parameters and a splat argument. [Feature #10440] (Details are described in this ticket) Most of complex part is moved to vm_args.c. Now, ISeq#to_a does not catch up new instruction format. * vm_core.h: change iseq data structures. * introduce rb_call_info_kw_arg_t to represent keyword arguments. * add rb_call_info_t::kw_arg. * rename rb_iseq_t::arg_post_len to rb_iseq_t::arg_post_num. * rename rb_iseq_t::arg_keywords to arg_keyword_num. * rename rb_iseq_t::arg_keyword to rb_iseq_t::arg_keyword_bits. to represent keyword bitmap parameter index. This bitmap parameter shows that which keyword parameters are given or not given (0 for given). It is refered by `checkkeyword' instruction described bellow. * rename rb_iseq_t::arg_keyword_check to rb_iseq_t::arg_keyword_rest to represent keyword rest parameter index. * add rb_iseq_t::arg_keyword_default_values to represent default keyword values. * rename VM_CALL_ARGS_SKIP_SETUP to VM_CALL_ARGS_SIMPLE to represent (ci->flag & (SPLAT|BLOCKARG)) && ci->blockiseq == NULL && ci->kw_arg == NULL. * vm_insnhelper.c, vm_args.c: rewrite with refactoring. * rewrite splat argument code. * rewrite keyword arguments/parameters code. * merge method and block parameter fitting code into one code base. * vm.c, vm_eval.c: catch up these changes. * compile.c (new_callinfo): callinfo requires kw_arg parameter. * compile.c (compile_array_): check the last argument Hash object or not. If Hash object and all keys are Symbol literals, they are compiled to keyword arguments. * insns.def (checkkeyword): add new instruction. This instruction check the availability of corresponding keyword. For example, a method "def foo k1: 'v1'; end" is cimpiled to the following instructions. 0000 checkkeyword 2, 0 # check k1 is given. 0003 branchif 9 # if given, jump to address #9 0005 putstring "v1" 0007 setlocal_OP__WC__0 3 # k1 = 'v1' 0009 trace 8 0011 putnil 0012 trace 16 0014 leave * insns.def (opt_send_simple): removed and add new instruction "opt_send_without_block". * parse.y (new_args_tail_gen): reorder variables. Before this patch, a method "def foo(k1: 1, kr1:, k2: 2, **krest, &b)" has parameter variables "k1, kr1, k2, &b, internal_id, krest", but this patch reorders to "kr1, k1, k2, internal_id, krest, &b". (locate a block variable at last) * parse.y (vtable_pop): added. This function remove latest `n' variables from vtable. * iseq.c: catch up iseq data changes. * proc.c: ditto. * class.c (keyword_error): export as rb_keyword_error(). * common.mk: depend vm_args.c for vm.o. * hash.c (rb_hash_has_key): export. * internal.h: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@48239 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2014-11-02 21:02:55 +03:00
if (local_body->param.flags.has_kwrest) {
int idx = local_body->local_table_size - local_kwd->rest_start;
ADD_GETLOCAL(args, line, idx, lvar_level);
* rewrite method/block parameter fitting logic to optimize keyword arguments/parameters and a splat argument. [Feature #10440] (Details are described in this ticket) Most of complex part is moved to vm_args.c. Now, ISeq#to_a does not catch up new instruction format. * vm_core.h: change iseq data structures. * introduce rb_call_info_kw_arg_t to represent keyword arguments. * add rb_call_info_t::kw_arg. * rename rb_iseq_t::arg_post_len to rb_iseq_t::arg_post_num. * rename rb_iseq_t::arg_keywords to arg_keyword_num. * rename rb_iseq_t::arg_keyword to rb_iseq_t::arg_keyword_bits. to represent keyword bitmap parameter index. This bitmap parameter shows that which keyword parameters are given or not given (0 for given). It is refered by `checkkeyword' instruction described bellow. * rename rb_iseq_t::arg_keyword_check to rb_iseq_t::arg_keyword_rest to represent keyword rest parameter index. * add rb_iseq_t::arg_keyword_default_values to represent default keyword values. * rename VM_CALL_ARGS_SKIP_SETUP to VM_CALL_ARGS_SIMPLE to represent (ci->flag & (SPLAT|BLOCKARG)) && ci->blockiseq == NULL && ci->kw_arg == NULL. * vm_insnhelper.c, vm_args.c: rewrite with refactoring. * rewrite splat argument code. * rewrite keyword arguments/parameters code. * merge method and block parameter fitting code into one code base. * vm.c, vm_eval.c: catch up these changes. * compile.c (new_callinfo): callinfo requires kw_arg parameter. * compile.c (compile_array_): check the last argument Hash object or not. If Hash object and all keys are Symbol literals, they are compiled to keyword arguments. * insns.def (checkkeyword): add new instruction. This instruction check the availability of corresponding keyword. For example, a method "def foo k1: 'v1'; end" is cimpiled to the following instructions. 0000 checkkeyword 2, 0 # check k1 is given. 0003 branchif 9 # if given, jump to address #9 0005 putstring "v1" 0007 setlocal_OP__WC__0 3 # k1 = 'v1' 0009 trace 8 0011 putnil 0012 trace 16 0014 leave * insns.def (opt_send_simple): removed and add new instruction "opt_send_without_block". * parse.y (new_args_tail_gen): reorder variables. Before this patch, a method "def foo(k1: 1, kr1:, k2: 2, **krest, &b)" has parameter variables "k1, kr1, k2, &b, internal_id, krest", but this patch reorders to "kr1, k1, k2, internal_id, krest, &b". (locate a block variable at last) * parse.y (vtable_pop): added. This function remove latest `n' variables from vtable. * iseq.c: catch up iseq data changes. * proc.c: ditto. * class.c (keyword_error): export as rb_keyword_error(). * common.mk: depend vm_args.c for vm.o. * hash.c (rb_hash_has_key): export. * internal.h: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@48239 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2014-11-02 21:02:55 +03:00
ADD_SEND (args, line, rb_intern("dup"), INT2FIX(0));
}
else {
ADD_INSN1(args, line, newhash, INT2FIX(0));
}
for (i = 0; i < local_kwd->num; ++i) {
ID id = local_kwd->table[i];
int idx = local_size - get_local_var_idx(liseq, id);
ADD_INSN1(args, line, putobject, ID2SYM(id));
ADD_GETLOCAL(args, line, idx, lvar_level);
}
ADD_SEND(args, line, id_core_hash_merge_ptr, INT2FIX(i * 2 + 1));
if (local_body->param.flags.has_rest) {
ADD_INSN1(args, line, newarray, INT2FIX(1));
ADD_INSN (args, line, concatarray);
--argc;
}
flag |= VM_CALL_KW_SPLAT;
}
else if (local_body->param.flags.has_kwrest) {
int idx = local_body->local_table_size - local_kwd->rest_start;
ADD_GETLOCAL(args, line, idx, lvar_level);
ADD_SEND (args, line, rb_intern("dup"), INT2FIX(0));
if (local_body->param.flags.has_rest) {
ADD_INSN1(args, line, newarray, INT2FIX(1));
ADD_INSN (args, line, concatarray);
}
else {
argc++;
* rewrite method/block parameter fitting logic to optimize keyword arguments/parameters and a splat argument. [Feature #10440] (Details are described in this ticket) Most of complex part is moved to vm_args.c. Now, ISeq#to_a does not catch up new instruction format. * vm_core.h: change iseq data structures. * introduce rb_call_info_kw_arg_t to represent keyword arguments. * add rb_call_info_t::kw_arg. * rename rb_iseq_t::arg_post_len to rb_iseq_t::arg_post_num. * rename rb_iseq_t::arg_keywords to arg_keyword_num. * rename rb_iseq_t::arg_keyword to rb_iseq_t::arg_keyword_bits. to represent keyword bitmap parameter index. This bitmap parameter shows that which keyword parameters are given or not given (0 for given). It is refered by `checkkeyword' instruction described bellow. * rename rb_iseq_t::arg_keyword_check to rb_iseq_t::arg_keyword_rest to represent keyword rest parameter index. * add rb_iseq_t::arg_keyword_default_values to represent default keyword values. * rename VM_CALL_ARGS_SKIP_SETUP to VM_CALL_ARGS_SIMPLE to represent (ci->flag & (SPLAT|BLOCKARG)) && ci->blockiseq == NULL && ci->kw_arg == NULL. * vm_insnhelper.c, vm_args.c: rewrite with refactoring. * rewrite splat argument code. * rewrite keyword arguments/parameters code. * merge method and block parameter fitting code into one code base. * vm.c, vm_eval.c: catch up these changes. * compile.c (new_callinfo): callinfo requires kw_arg parameter. * compile.c (compile_array_): check the last argument Hash object or not. If Hash object and all keys are Symbol literals, they are compiled to keyword arguments. * insns.def (checkkeyword): add new instruction. This instruction check the availability of corresponding keyword. For example, a method "def foo k1: 'v1'; end" is cimpiled to the following instructions. 0000 checkkeyword 2, 0 # check k1 is given. 0003 branchif 9 # if given, jump to address #9 0005 putstring "v1" 0007 setlocal_OP__WC__0 3 # k1 = 'v1' 0009 trace 8 0011 putnil 0012 trace 16 0014 leave * insns.def (opt_send_simple): removed and add new instruction "opt_send_without_block". * parse.y (new_args_tail_gen): reorder variables. Before this patch, a method "def foo(k1: 1, kr1:, k2: 2, **krest, &b)" has parameter variables "k1, kr1, k2, &b, internal_id, krest", but this patch reorders to "kr1, k1, k2, internal_id, krest, &b". (locate a block variable at last) * parse.y (vtable_pop): added. This function remove latest `n' variables from vtable. * iseq.c: catch up iseq data changes. * proc.c: ditto. * class.c (keyword_error): export as rb_keyword_error(). * common.mk: depend vm_args.c for vm.o. * hash.c (rb_hash_has_key): export. * internal.h: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@48239 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2014-11-02 21:02:55 +03:00
}
flag |= VM_CALL_KW_SPLAT;
}
}
ADD_INSN(ret, line, putself);
ADD_SEQ(ret, args);
ADD_INSN2(ret, line, invokesuper,
new_callinfo(iseq, 0, argc, flag | VM_CALL_SUPER | (type == NODE_ZSUPER ? VM_CALL_ZSUPER : 0) | VM_CALL_FCALL, keywords, parent_block != NULL),
parent_block);
if (popped) {
ADD_INSN(ret, line, pop);
}
break;
}
case NODE_LIST:{
CHECK(compile_array(iseq, ret, node, popped) >= 0);
break;
}
case NODE_ZLIST:{
if (!popped) {
ADD_INSN1(ret, line, newarray, INT2FIX(0));
}
break;
}
case NODE_VALUES:{
const NODE *n = node;
if (popped) {
COMPILE_ERROR(ERROR_ARGS "NODE_VALUES: must not be popped");
}
while (n) {
CHECK(COMPILE(ret, "values item", n->nd_head));
n = n->nd_next;
}
ADD_INSN1(ret, line, newarray, INT2FIX(node->nd_alen));
break;
}
case NODE_HASH:
CHECK(compile_hash(iseq, ret, node, popped) >= 0);
break;
case NODE_RETURN:
CHECK(compile_return(iseq, ret, node, popped));
break;
case NODE_YIELD:{
DECL_ANCHOR(args);
VALUE argc;
unsigned int flag = 0;
struct rb_call_info_kw_arg *keywords = NULL;
INIT_ANCHOR(args);
if (check_yield_place(iseq, line) == FALSE) {
COMPILE_ERROR(ERROR_ARGS "Invalid yield");
goto ng;
}
if (node->nd_head) {
* rewrite method/block parameter fitting logic to optimize keyword arguments/parameters and a splat argument. [Feature #10440] (Details are described in this ticket) Most of complex part is moved to vm_args.c. Now, ISeq#to_a does not catch up new instruction format. * vm_core.h: change iseq data structures. * introduce rb_call_info_kw_arg_t to represent keyword arguments. * add rb_call_info_t::kw_arg. * rename rb_iseq_t::arg_post_len to rb_iseq_t::arg_post_num. * rename rb_iseq_t::arg_keywords to arg_keyword_num. * rename rb_iseq_t::arg_keyword to rb_iseq_t::arg_keyword_bits. to represent keyword bitmap parameter index. This bitmap parameter shows that which keyword parameters are given or not given (0 for given). It is refered by `checkkeyword' instruction described bellow. * rename rb_iseq_t::arg_keyword_check to rb_iseq_t::arg_keyword_rest to represent keyword rest parameter index. * add rb_iseq_t::arg_keyword_default_values to represent default keyword values. * rename VM_CALL_ARGS_SKIP_SETUP to VM_CALL_ARGS_SIMPLE to represent (ci->flag & (SPLAT|BLOCKARG)) && ci->blockiseq == NULL && ci->kw_arg == NULL. * vm_insnhelper.c, vm_args.c: rewrite with refactoring. * rewrite splat argument code. * rewrite keyword arguments/parameters code. * merge method and block parameter fitting code into one code base. * vm.c, vm_eval.c: catch up these changes. * compile.c (new_callinfo): callinfo requires kw_arg parameter. * compile.c (compile_array_): check the last argument Hash object or not. If Hash object and all keys are Symbol literals, they are compiled to keyword arguments. * insns.def (checkkeyword): add new instruction. This instruction check the availability of corresponding keyword. For example, a method "def foo k1: 'v1'; end" is cimpiled to the following instructions. 0000 checkkeyword 2, 0 # check k1 is given. 0003 branchif 9 # if given, jump to address #9 0005 putstring "v1" 0007 setlocal_OP__WC__0 3 # k1 = 'v1' 0009 trace 8 0011 putnil 0012 trace 16 0014 leave * insns.def (opt_send_simple): removed and add new instruction "opt_send_without_block". * parse.y (new_args_tail_gen): reorder variables. Before this patch, a method "def foo(k1: 1, kr1:, k2: 2, **krest, &b)" has parameter variables "k1, kr1, k2, &b, internal_id, krest", but this patch reorders to "kr1, k1, k2, internal_id, krest, &b". (locate a block variable at last) * parse.y (vtable_pop): added. This function remove latest `n' variables from vtable. * iseq.c: catch up iseq data changes. * proc.c: ditto. * class.c (keyword_error): export as rb_keyword_error(). * common.mk: depend vm_args.c for vm.o. * hash.c (rb_hash_has_key): export. * internal.h: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@48239 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2014-11-02 21:02:55 +03:00
argc = setup_args(iseq, args, node->nd_head, &flag, &keywords);
CHECK(!NIL_P(argc));
}
else {
argc = INT2FIX(0);
}
ADD_SEQ(ret, args);
ADD_INSN1(ret, line, invokeblock, new_callinfo(iseq, 0, FIX2INT(argc), flag, keywords, FALSE));
if (popped) {
ADD_INSN(ret, line, pop);
}
break;
}
case NODE_LVAR:{
if (!popped) {
ID id = node->nd_vid;
int idx = body->local_iseq->body->local_table_size - get_local_var_idx(iseq, id);
debugs("id: %s idx: %d\n", rb_id2name(id), idx);
ADD_GETLOCAL(ret, line, idx, get_lvar_level(iseq));
}
break;
}
case NODE_DVAR:{
int lv, idx, ls;
debugi("nd_vid", node->nd_vid);
if (!popped) {
idx = get_dyna_var_idx(iseq, node->nd_vid, &lv, &ls);
if (idx < 0) {
COMPILE_ERROR(ERROR_ARGS "unknown dvar (%"PRIsVALUE")",
rb_id2str(node->nd_vid));
goto ng;
}
ADD_GETLOCAL(ret, line, ls - idx, lv);
}
break;
}
case NODE_GVAR:{
ADD_INSN1(ret, line, getglobal,
((VALUE)node->nd_entry | 1));
if (popped) {
ADD_INSN(ret, line, pop);
}
break;
}
case NODE_IVAR:{
debugi("nd_vid", node->nd_vid);
if (!popped) {
ADD_INSN2(ret, line, getinstancevariable,
ID2SYM(node->nd_vid),
get_ivar_ic_value(iseq,node->nd_vid));
}
break;
}
case NODE_CONST:{
debugi("nd_vid", node->nd_vid);
if (ISEQ_COMPILE_DATA(iseq)->option->inline_const_cache) {
LABEL *lend = NEW_LABEL(line);
int ic_index = body->is_size++;
ADD_INSN2(ret, line, opt_getinlinecache, lend, INT2FIX(ic_index));
ADD_INSN1(ret, line, putobject, Qtrue);
ADD_INSN1(ret, line, getconstant, ID2SYM(node->nd_vid));
ADD_INSN1(ret, line, opt_setinlinecache, INT2FIX(ic_index));
ADD_LABEL(ret, lend);
}
else {
ADD_INSN(ret, line, putnil);
ADD_INSN1(ret, line, putobject, Qtrue);
ADD_INSN1(ret, line, getconstant, ID2SYM(node->nd_vid));
}
if (popped) {
ADD_INSN(ret, line, pop);
}
break;
}
case NODE_CVAR:{
if (!popped) {
ADD_INSN1(ret, line, getclassvariable,
ID2SYM(node->nd_vid));
}
break;
}
case NODE_NTH_REF:{
if (!popped) {
if (!node->nd_nth) {
ADD_INSN(ret, line, putnil);
break;
}
ADD_INSN2(ret, line, getspecial, INT2FIX(1) /* '~' */,
INT2FIX(node->nd_nth << 1));
}
break;
}
case NODE_BACK_REF:{
if (!popped) {
ADD_INSN2(ret, line, getspecial, INT2FIX(1) /* '~' */,
INT2FIX(0x01 | (node->nd_nth << 1)));
}
break;
}
case NODE_MATCH:
case NODE_MATCH2:
case NODE_MATCH3:{
DECL_ANCHOR(recv);
DECL_ANCHOR(val);
INIT_ANCHOR(recv);
INIT_ANCHOR(val);
switch (nd_type(node)) {
case NODE_MATCH:
ADD_INSN1(recv, line, putobject, node->nd_lit);
ADD_INSN2(val, line, getspecial, INT2FIX(0),
INT2FIX(0));
break;
case NODE_MATCH2:
CHECK(COMPILE(recv, "receiver", node->nd_recv));
CHECK(COMPILE(val, "value", node->nd_value));
break;
case NODE_MATCH3:
CHECK(COMPILE(recv, "receiver", node->nd_value));
CHECK(COMPILE(val, "value", node->nd_recv));
break;
}
ADD_SEQ(ret, recv);
ADD_SEQ(ret, val);
ADD_SEND(ret, line, idEqTilde, INT2FIX(1));
if (node->nd_args) {
compile_named_capture_assign(iseq, ret, node->nd_args);
}
if (popped) {
ADD_INSN(ret, line, pop);
}
break;
}
case NODE_LIT:{
debugp_param("lit", node->nd_lit);
if (!popped) {
ADD_INSN1(ret, line, putobject, node->nd_lit);
}
break;
}
case NODE_STR:{
debugp_param("nd_lit", node->nd_lit);
if (!popped) {
VALUE lit = node->nd_lit;
if (!ISEQ_COMPILE_DATA(iseq)->option->frozen_string_literal) {
lit = rb_fstring(lit);
ADD_INSN1(ret, line, putstring, lit);
RB_OBJ_WRITTEN(iseq, Qundef, lit);
}
else {
if (ISEQ_COMPILE_DATA(iseq)->option->debug_frozen_string_literal || RTEST(ruby_debug)) {
VALUE debug_info = rb_ary_new_from_args(2, rb_iseq_path(iseq), INT2FIX(line));
lit = rb_str_dup(lit);
rb_ivar_set(lit, id_debug_created_info, rb_obj_freeze(debug_info));
lit = rb_str_freeze(lit);
}
else {
lit = rb_fstring(lit);
}
ADD_INSN1(ret, line, putobject, lit);
RB_OBJ_WRITTEN(iseq, Qundef, lit);
}
}
break;
}
case NODE_DSTR:{
compile_dstr(iseq, ret, node);
if (popped) {
ADD_INSN(ret, line, pop);
}
else {
if (ISEQ_COMPILE_DATA(iseq)->option->frozen_string_literal) {
VALUE debug_info = Qnil;
if (ISEQ_COMPILE_DATA(iseq)->option->debug_frozen_string_literal || RTEST(ruby_debug)) {
debug_info = rb_ary_new_from_args(2, rb_iseq_path(iseq), INT2FIX(line));
}
ADD_INSN1(ret, line, freezestring, debug_info);
if (!NIL_P(debug_info)) {
RB_OBJ_WRITTEN(iseq, Qundef, rb_obj_freeze(debug_info));
}
}
}
break;
}
case NODE_XSTR:{
ADD_CALL_RECEIVER(ret, line);
VALUE str = rb_fstring(node->nd_lit);
ADD_INSN1(ret, line, putobject, str);
RB_OBJ_WRITTEN(iseq, Qundef, str);
ADD_CALL(ret, line, idBackquote, INT2FIX(1));
if (popped) {
ADD_INSN(ret, line, pop);
}
break;
}
case NODE_DXSTR:{
ADD_CALL_RECEIVER(ret, line);
compile_dstr(iseq, ret, node);
ADD_CALL(ret, line, idBackquote, INT2FIX(1));
if (popped) {
ADD_INSN(ret, line, pop);
}
break;
}
case NODE_EVSTR:
CHECK(compile_evstr(iseq, ret, node->nd_body, popped));
break;
case NODE_DREGX:{
compile_dregx(iseq, ret, node);
if (popped) {
ADD_INSN(ret, line, pop);
}
break;
}
case NODE_ONCE:{
int ic_index = body->is_size++;
const rb_iseq_t *block_iseq;
block_iseq = NEW_CHILD_ISEQ(node->nd_body, make_name_for_block(iseq), ISEQ_TYPE_PLAIN, line);
ADD_INSN2(ret, line, once, block_iseq, INT2FIX(ic_index));
RB_OBJ_WRITTEN(iseq, Qundef, (VALUE)block_iseq);
if (popped) {
ADD_INSN(ret, line, pop);
}
break;
}
case NODE_ARGSCAT:{
if (popped) {
CHECK(COMPILE(ret, "argscat head", node->nd_head));
ADD_INSN1(ret, line, splatarray, Qfalse);
ADD_INSN(ret, line, pop);
CHECK(COMPILE(ret, "argscat body", node->nd_body));
ADD_INSN1(ret, line, splatarray, Qfalse);
ADD_INSN(ret, line, pop);
}
else {
CHECK(COMPILE(ret, "argscat head", node->nd_head));
CHECK(COMPILE(ret, "argscat body", node->nd_body));
ADD_INSN(ret, line, concatarray);
}
break;
}
case NODE_ARGSPUSH:{
if (popped) {
CHECK(COMPILE(ret, "arsgpush head", node->nd_head));
ADD_INSN1(ret, line, splatarray, Qfalse);
ADD_INSN(ret, line, pop);
CHECK(COMPILE_(ret, "argspush body", node->nd_body, popped));
}
else {
CHECK(COMPILE(ret, "arsgpush head", node->nd_head));
CHECK(COMPILE_(ret, "argspush body", node->nd_body, popped));
ADD_INSN1(ret, line, newarray, INT2FIX(1));
ADD_INSN(ret, line, concatarray);
}
break;
}
case NODE_SPLAT:{
CHECK(COMPILE(ret, "splat", node->nd_head));
ADD_INSN1(ret, line, splatarray, Qtrue);
if (popped) {
ADD_INSN(ret, line, pop);
}
break;
}
case NODE_DEFN:{
ID mid = node->nd_mid;
2015-07-22 01:52:59 +03:00
const rb_iseq_t *method_iseq = NEW_ISEQ(node->nd_defn,
rb_id2str(mid),
2015-07-22 01:52:59 +03:00
ISEQ_TYPE_METHOD, line);
debugp_param("defn/iseq", rb_iseqw_new(method_iseq));
ADD_INSN2(ret, line, definemethod, ID2SYM(mid), method_iseq);
RB_OBJ_WRITTEN(iseq, Qundef, (VALUE)method_iseq);
if (!popped) {
ADD_INSN1(ret, line, putobject, ID2SYM(mid));
}
break;
}
case NODE_DEFS:{
ID mid = node->nd_mid;
const rb_iseq_t * singleton_method_iseq = NEW_ISEQ(node->nd_defn,
rb_id2str(mid),
ISEQ_TYPE_METHOD, line);
debugp_param("defs/iseq", rb_iseqw_new(singleton_method_iseq));
CHECK(COMPILE(ret, "defs: recv", node->nd_recv));
ADD_INSN2(ret, line, definesmethod, ID2SYM(mid), singleton_method_iseq);
RB_OBJ_WRITTEN(iseq, Qundef, (VALUE)singleton_method_iseq);
if (!popped) {
ADD_INSN1(ret, line, putobject, ID2SYM(mid));
}
break;
}
case NODE_ALIAS:{
ADD_INSN1(ret, line, putspecialobject, INT2FIX(VM_SPECIAL_OBJECT_VMCORE));
ADD_INSN1(ret, line, putspecialobject, INT2FIX(VM_SPECIAL_OBJECT_CBASE));
CHECK(COMPILE(ret, "alias arg1", node->nd_1st));
CHECK(COMPILE(ret, "alias arg2", node->nd_2nd));
ADD_SEND(ret, line, id_core_set_method_alias, INT2FIX(3));
if (popped) {
ADD_INSN(ret, line, pop);
}
break;
}
case NODE_VALIAS:{
ADD_INSN1(ret, line, putspecialobject, INT2FIX(VM_SPECIAL_OBJECT_VMCORE));
ADD_INSN1(ret, line, putobject, ID2SYM(node->nd_alias));
ADD_INSN1(ret, line, putobject, ID2SYM(node->nd_orig));
ADD_SEND(ret, line, id_core_set_variable_alias, INT2FIX(2));
if (popped) {
ADD_INSN(ret, line, pop);
}
break;
}
case NODE_UNDEF:{
ADD_INSN1(ret, line, putspecialobject, INT2FIX(VM_SPECIAL_OBJECT_VMCORE));
ADD_INSN1(ret, line, putspecialobject, INT2FIX(VM_SPECIAL_OBJECT_CBASE));
CHECK(COMPILE(ret, "undef arg", node->nd_undef));
ADD_SEND(ret, line, id_core_undef_method, INT2FIX(2));
if (popped) {
ADD_INSN(ret, line, pop);
}
break;
}
case NODE_CLASS:{
2015-07-22 01:52:59 +03:00
const rb_iseq_t *class_iseq = NEW_CHILD_ISEQ(node->nd_body,
rb_sprintf("<class:%"PRIsVALUE">", rb_id2str(node->nd_cpath->nd_mid)),
ISEQ_TYPE_CLASS, line);
const int flags = VM_DEFINECLASS_TYPE_CLASS |
(node->nd_super ? VM_DEFINECLASS_FLAG_HAS_SUPERCLASS : 0) |
compile_cpath(ret, iseq, node->nd_cpath);
2015-07-22 01:52:59 +03:00
CHECK(COMPILE(ret, "super", node->nd_super));
2015-07-22 01:52:59 +03:00
ADD_INSN3(ret, line, defineclass, ID2SYM(node->nd_cpath->nd_mid), class_iseq, INT2FIX(flags));
RB_OBJ_WRITTEN(iseq, Qundef, (VALUE)class_iseq);
if (popped) {
ADD_INSN(ret, line, pop);
}
break;
}
case NODE_MODULE:{
2015-07-22 01:52:59 +03:00
const rb_iseq_t *module_iseq = NEW_CHILD_ISEQ(node->nd_body,
rb_sprintf("<module:%"PRIsVALUE">", rb_id2str(node->nd_cpath->nd_mid)),
ISEQ_TYPE_CLASS, line);
const int flags = VM_DEFINECLASS_TYPE_MODULE |
compile_cpath(ret, iseq, node->nd_cpath);
2015-07-22 01:52:59 +03:00
ADD_INSN (ret, line, putnil); /* dummy */
2015-07-22 01:52:59 +03:00
ADD_INSN3(ret, line, defineclass, ID2SYM(node->nd_cpath->nd_mid), module_iseq, INT2FIX(flags));
RB_OBJ_WRITTEN(iseq, Qundef, (VALUE)module_iseq);
2015-07-22 01:52:59 +03:00
if (popped) {
ADD_INSN(ret, line, pop);
}
break;
}
case NODE_SCLASS:{
ID singletonclass;
const rb_iseq_t *singleton_class = NEW_ISEQ(node->nd_body, rb_fstring_lit("singleton class"),
2015-07-22 01:52:59 +03:00
ISEQ_TYPE_CLASS, line);
CHECK(COMPILE(ret, "sclass#recv", node->nd_recv));
ADD_INSN (ret, line, putnil);
CONST_ID(singletonclass, "singletonclass");
ADD_INSN3(ret, line, defineclass,
2015-07-22 01:52:59 +03:00
ID2SYM(singletonclass), singleton_class,
INT2FIX(VM_DEFINECLASS_TYPE_SINGLETON_CLASS));
RB_OBJ_WRITTEN(iseq, Qundef, (VALUE)singleton_class);
if (popped) {
ADD_INSN(ret, line, pop);
}
break;
}
case NODE_COLON2:{
if (rb_is_const_id(node->nd_mid)) {
/* constant */
LABEL *lend = NEW_LABEL(line);
int ic_index = body->is_size++;
DECL_ANCHOR(pref);
DECL_ANCHOR(body);
INIT_ANCHOR(pref);
INIT_ANCHOR(body);
CHECK(compile_const_prefix(iseq, node, pref, body));
if (LIST_INSN_SIZE_ZERO(pref)) {
if (ISEQ_COMPILE_DATA(iseq)->option->inline_const_cache) {
ADD_INSN2(ret, line, opt_getinlinecache, lend, INT2FIX(ic_index));
}
else {
ADD_INSN(ret, line, putnil);
}
ADD_SEQ(ret, body);
if (ISEQ_COMPILE_DATA(iseq)->option->inline_const_cache) {
ADD_INSN1(ret, line, opt_setinlinecache, INT2FIX(ic_index));
ADD_LABEL(ret, lend);
}
}
else {
ADD_SEQ(ret, pref);
ADD_SEQ(ret, body);
}
}
else {
/* function call */
ADD_CALL_RECEIVER(ret, line);
CHECK(COMPILE(ret, "colon2#nd_head", node->nd_head));
* rewrite method/block parameter fitting logic to optimize keyword arguments/parameters and a splat argument. [Feature #10440] (Details are described in this ticket) Most of complex part is moved to vm_args.c. Now, ISeq#to_a does not catch up new instruction format. * vm_core.h: change iseq data structures. * introduce rb_call_info_kw_arg_t to represent keyword arguments. * add rb_call_info_t::kw_arg. * rename rb_iseq_t::arg_post_len to rb_iseq_t::arg_post_num. * rename rb_iseq_t::arg_keywords to arg_keyword_num. * rename rb_iseq_t::arg_keyword to rb_iseq_t::arg_keyword_bits. to represent keyword bitmap parameter index. This bitmap parameter shows that which keyword parameters are given or not given (0 for given). It is refered by `checkkeyword' instruction described bellow. * rename rb_iseq_t::arg_keyword_check to rb_iseq_t::arg_keyword_rest to represent keyword rest parameter index. * add rb_iseq_t::arg_keyword_default_values to represent default keyword values. * rename VM_CALL_ARGS_SKIP_SETUP to VM_CALL_ARGS_SIMPLE to represent (ci->flag & (SPLAT|BLOCKARG)) && ci->blockiseq == NULL && ci->kw_arg == NULL. * vm_insnhelper.c, vm_args.c: rewrite with refactoring. * rewrite splat argument code. * rewrite keyword arguments/parameters code. * merge method and block parameter fitting code into one code base. * vm.c, vm_eval.c: catch up these changes. * compile.c (new_callinfo): callinfo requires kw_arg parameter. * compile.c (compile_array_): check the last argument Hash object or not. If Hash object and all keys are Symbol literals, they are compiled to keyword arguments. * insns.def (checkkeyword): add new instruction. This instruction check the availability of corresponding keyword. For example, a method "def foo k1: 'v1'; end" is cimpiled to the following instructions. 0000 checkkeyword 2, 0 # check k1 is given. 0003 branchif 9 # if given, jump to address #9 0005 putstring "v1" 0007 setlocal_OP__WC__0 3 # k1 = 'v1' 0009 trace 8 0011 putnil 0012 trace 16 0014 leave * insns.def (opt_send_simple): removed and add new instruction "opt_send_without_block". * parse.y (new_args_tail_gen): reorder variables. Before this patch, a method "def foo(k1: 1, kr1:, k2: 2, **krest, &b)" has parameter variables "k1, kr1, k2, &b, internal_id, krest", but this patch reorders to "kr1, k1, k2, internal_id, krest, &b". (locate a block variable at last) * parse.y (vtable_pop): added. This function remove latest `n' variables from vtable. * iseq.c: catch up iseq data changes. * proc.c: ditto. * class.c (keyword_error): export as rb_keyword_error(). * common.mk: depend vm_args.c for vm.o. * hash.c (rb_hash_has_key): export. * internal.h: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@48239 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2014-11-02 21:02:55 +03:00
ADD_CALL(ret, line, node->nd_mid, INT2FIX(1));
}
if (popped) {
ADD_INSN(ret, line, pop);
}
break;
}
case NODE_COLON3:{
LABEL *lend = NEW_LABEL(line);
int ic_index = body->is_size++;
debugi("colon3#nd_mid", node->nd_mid);
/* add cache insn */
if (ISEQ_COMPILE_DATA(iseq)->option->inline_const_cache) {
ADD_INSN2(ret, line, opt_getinlinecache, lend, INT2FIX(ic_index));
ADD_INSN(ret, line, pop);
}
ADD_INSN1(ret, line, putobject, rb_cObject);
ADD_INSN1(ret, line, putobject, Qtrue);
ADD_INSN1(ret, line, getconstant, ID2SYM(node->nd_mid));
if (ISEQ_COMPILE_DATA(iseq)->option->inline_const_cache) {
ADD_INSN1(ret, line, opt_setinlinecache, INT2FIX(ic_index));
ADD_LABEL(ret, lend);
}
if (popped) {
ADD_INSN(ret, line, pop);
}
break;
}
case NODE_DOT2:
case NODE_DOT3:{
int excl = type == NODE_DOT3;
VALUE flag = INT2FIX(excl);
const NODE *b = node->nd_beg;
const NODE *e = node->nd_end;
if (number_literal_p(b) && number_literal_p(e)) {
if (!popped) {
VALUE val = rb_range_new(b->nd_lit, e->nd_lit, excl);
ADD_INSN1(ret, line, putobject, val);
RB_OBJ_WRITTEN(iseq, Qundef, val);
}
}
else {
CHECK(COMPILE_(ret, "min", b, popped));
CHECK(COMPILE_(ret, "max", e, popped));
if (!popped) {
ADD_INSN1(ret, line, newrange, flag);
}
}
break;
}
case NODE_FLIP2:
case NODE_FLIP3:{
LABEL *lend = NEW_LABEL(line);
LABEL *ltrue = NEW_LABEL(line);
LABEL *lfalse = NEW_LABEL(line);
CHECK(compile_flip_flop(iseq, ret, node, type == NODE_FLIP2,
ltrue, lfalse));
ADD_LABEL(ret, ltrue);
ADD_INSN1(ret, line, putobject, Qtrue);
ADD_INSNL(ret, line, jump, lend);
ADD_LABEL(ret, lfalse);
ADD_INSN1(ret, line, putobject, Qfalse);
ADD_LABEL(ret, lend);
break;
}
case NODE_SELF:{
if (!popped) {
ADD_INSN(ret, line, putself);
}
break;
}
case NODE_NIL:{
if (!popped) {
ADD_INSN(ret, line, putnil);
}
break;
}
case NODE_TRUE:{
if (!popped) {
ADD_INSN1(ret, line, putobject, Qtrue);
}
break;
}
case NODE_FALSE:{
if (!popped) {
ADD_INSN1(ret, line, putobject, Qfalse);
}
break;
}
case NODE_ERRINFO:{
if (!popped) {
if (body->type == ISEQ_TYPE_RESCUE) {
ADD_GETLOCAL(ret, line, LVAR_ERRINFO, 0);
}
else {
const rb_iseq_t *ip = iseq;
int level = 0;
while (ip) {
2015-07-22 01:52:59 +03:00
if (ip->body->type == ISEQ_TYPE_RESCUE) {
break;
}
2015-07-22 01:52:59 +03:00
ip = ip->body->parent_iseq;
level++;
}
if (ip) {
ADD_GETLOCAL(ret, line, LVAR_ERRINFO, level);
}
else {
ADD_INSN(ret, line, putnil);
}
}
}
break;
}
case NODE_DEFINED:
if (!popped) {
CHECK(compile_defined_expr(iseq, ret, node, Qtrue));
}
break;
case NODE_POSTEXE:{
/* compiled to:
* ONCE{ rb_mRubyVMFrozenCore::core#set_postexe{ ... } }
*/
int is_index = body->is_size++;
struct rb_iseq_new_with_callback_callback_func *ifunc =
rb_iseq_new_with_callback_new_callback(build_postexe_iseq, node->nd_body);
const rb_iseq_t *once_iseq =
new_child_iseq_with_callback(iseq, ifunc,
rb_fstring(make_name_for_block(iseq)), iseq, ISEQ_TYPE_BLOCK, line);
ADD_INSN2(ret, line, once, once_iseq, INT2FIX(is_index));
RB_OBJ_WRITTEN(iseq, Qundef, (VALUE)once_iseq);
if (popped) {
ADD_INSN(ret, line, pop);
}
break;
}
* rewrite method/block parameter fitting logic to optimize keyword arguments/parameters and a splat argument. [Feature #10440] (Details are described in this ticket) Most of complex part is moved to vm_args.c. Now, ISeq#to_a does not catch up new instruction format. * vm_core.h: change iseq data structures. * introduce rb_call_info_kw_arg_t to represent keyword arguments. * add rb_call_info_t::kw_arg. * rename rb_iseq_t::arg_post_len to rb_iseq_t::arg_post_num. * rename rb_iseq_t::arg_keywords to arg_keyword_num. * rename rb_iseq_t::arg_keyword to rb_iseq_t::arg_keyword_bits. to represent keyword bitmap parameter index. This bitmap parameter shows that which keyword parameters are given or not given (0 for given). It is refered by `checkkeyword' instruction described bellow. * rename rb_iseq_t::arg_keyword_check to rb_iseq_t::arg_keyword_rest to represent keyword rest parameter index. * add rb_iseq_t::arg_keyword_default_values to represent default keyword values. * rename VM_CALL_ARGS_SKIP_SETUP to VM_CALL_ARGS_SIMPLE to represent (ci->flag & (SPLAT|BLOCKARG)) && ci->blockiseq == NULL && ci->kw_arg == NULL. * vm_insnhelper.c, vm_args.c: rewrite with refactoring. * rewrite splat argument code. * rewrite keyword arguments/parameters code. * merge method and block parameter fitting code into one code base. * vm.c, vm_eval.c: catch up these changes. * compile.c (new_callinfo): callinfo requires kw_arg parameter. * compile.c (compile_array_): check the last argument Hash object or not. If Hash object and all keys are Symbol literals, they are compiled to keyword arguments. * insns.def (checkkeyword): add new instruction. This instruction check the availability of corresponding keyword. For example, a method "def foo k1: 'v1'; end" is cimpiled to the following instructions. 0000 checkkeyword 2, 0 # check k1 is given. 0003 branchif 9 # if given, jump to address #9 0005 putstring "v1" 0007 setlocal_OP__WC__0 3 # k1 = 'v1' 0009 trace 8 0011 putnil 0012 trace 16 0014 leave * insns.def (opt_send_simple): removed and add new instruction "opt_send_without_block". * parse.y (new_args_tail_gen): reorder variables. Before this patch, a method "def foo(k1: 1, kr1:, k2: 2, **krest, &b)" has parameter variables "k1, kr1, k2, &b, internal_id, krest", but this patch reorders to "kr1, k1, k2, internal_id, krest, &b". (locate a block variable at last) * parse.y (vtable_pop): added. This function remove latest `n' variables from vtable. * iseq.c: catch up iseq data changes. * proc.c: ditto. * class.c (keyword_error): export as rb_keyword_error(). * common.mk: depend vm_args.c for vm.o. * hash.c (rb_hash_has_key): export. * internal.h: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@48239 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2014-11-02 21:02:55 +03:00
case NODE_KW_ARG:
{
LABEL *end_label = NEW_LABEL(nd_line(node));
const NODE *default_value = node->nd_body->nd_value;
* rewrite method/block parameter fitting logic to optimize keyword arguments/parameters and a splat argument. [Feature #10440] (Details are described in this ticket) Most of complex part is moved to vm_args.c. Now, ISeq#to_a does not catch up new instruction format. * vm_core.h: change iseq data structures. * introduce rb_call_info_kw_arg_t to represent keyword arguments. * add rb_call_info_t::kw_arg. * rename rb_iseq_t::arg_post_len to rb_iseq_t::arg_post_num. * rename rb_iseq_t::arg_keywords to arg_keyword_num. * rename rb_iseq_t::arg_keyword to rb_iseq_t::arg_keyword_bits. to represent keyword bitmap parameter index. This bitmap parameter shows that which keyword parameters are given or not given (0 for given). It is refered by `checkkeyword' instruction described bellow. * rename rb_iseq_t::arg_keyword_check to rb_iseq_t::arg_keyword_rest to represent keyword rest parameter index. * add rb_iseq_t::arg_keyword_default_values to represent default keyword values. * rename VM_CALL_ARGS_SKIP_SETUP to VM_CALL_ARGS_SIMPLE to represent (ci->flag & (SPLAT|BLOCKARG)) && ci->blockiseq == NULL && ci->kw_arg == NULL. * vm_insnhelper.c, vm_args.c: rewrite with refactoring. * rewrite splat argument code. * rewrite keyword arguments/parameters code. * merge method and block parameter fitting code into one code base. * vm.c, vm_eval.c: catch up these changes. * compile.c (new_callinfo): callinfo requires kw_arg parameter. * compile.c (compile_array_): check the last argument Hash object or not. If Hash object and all keys are Symbol literals, they are compiled to keyword arguments. * insns.def (checkkeyword): add new instruction. This instruction check the availability of corresponding keyword. For example, a method "def foo k1: 'v1'; end" is cimpiled to the following instructions. 0000 checkkeyword 2, 0 # check k1 is given. 0003 branchif 9 # if given, jump to address #9 0005 putstring "v1" 0007 setlocal_OP__WC__0 3 # k1 = 'v1' 0009 trace 8 0011 putnil 0012 trace 16 0014 leave * insns.def (opt_send_simple): removed and add new instruction "opt_send_without_block". * parse.y (new_args_tail_gen): reorder variables. Before this patch, a method "def foo(k1: 1, kr1:, k2: 2, **krest, &b)" has parameter variables "k1, kr1, k2, &b, internal_id, krest", but this patch reorders to "kr1, k1, k2, internal_id, krest, &b". (locate a block variable at last) * parse.y (vtable_pop): added. This function remove latest `n' variables from vtable. * iseq.c: catch up iseq data changes. * proc.c: ditto. * class.c (keyword_error): export as rb_keyword_error(). * common.mk: depend vm_args.c for vm.o. * hash.c (rb_hash_has_key): export. * internal.h: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@48239 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2014-11-02 21:02:55 +03:00
2019-06-04 17:17:38 +03:00
if (default_value == NODE_SPECIAL_REQUIRED_KEYWORD) {
* rewrite method/block parameter fitting logic to optimize keyword arguments/parameters and a splat argument. [Feature #10440] (Details are described in this ticket) Most of complex part is moved to vm_args.c. Now, ISeq#to_a does not catch up new instruction format. * vm_core.h: change iseq data structures. * introduce rb_call_info_kw_arg_t to represent keyword arguments. * add rb_call_info_t::kw_arg. * rename rb_iseq_t::arg_post_len to rb_iseq_t::arg_post_num. * rename rb_iseq_t::arg_keywords to arg_keyword_num. * rename rb_iseq_t::arg_keyword to rb_iseq_t::arg_keyword_bits. to represent keyword bitmap parameter index. This bitmap parameter shows that which keyword parameters are given or not given (0 for given). It is refered by `checkkeyword' instruction described bellow. * rename rb_iseq_t::arg_keyword_check to rb_iseq_t::arg_keyword_rest to represent keyword rest parameter index. * add rb_iseq_t::arg_keyword_default_values to represent default keyword values. * rename VM_CALL_ARGS_SKIP_SETUP to VM_CALL_ARGS_SIMPLE to represent (ci->flag & (SPLAT|BLOCKARG)) && ci->blockiseq == NULL && ci->kw_arg == NULL. * vm_insnhelper.c, vm_args.c: rewrite with refactoring. * rewrite splat argument code. * rewrite keyword arguments/parameters code. * merge method and block parameter fitting code into one code base. * vm.c, vm_eval.c: catch up these changes. * compile.c (new_callinfo): callinfo requires kw_arg parameter. * compile.c (compile_array_): check the last argument Hash object or not. If Hash object and all keys are Symbol literals, they are compiled to keyword arguments. * insns.def (checkkeyword): add new instruction. This instruction check the availability of corresponding keyword. For example, a method "def foo k1: 'v1'; end" is cimpiled to the following instructions. 0000 checkkeyword 2, 0 # check k1 is given. 0003 branchif 9 # if given, jump to address #9 0005 putstring "v1" 0007 setlocal_OP__WC__0 3 # k1 = 'v1' 0009 trace 8 0011 putnil 0012 trace 16 0014 leave * insns.def (opt_send_simple): removed and add new instruction "opt_send_without_block". * parse.y (new_args_tail_gen): reorder variables. Before this patch, a method "def foo(k1: 1, kr1:, k2: 2, **krest, &b)" has parameter variables "k1, kr1, k2, &b, internal_id, krest", but this patch reorders to "kr1, k1, k2, internal_id, krest, &b". (locate a block variable at last) * parse.y (vtable_pop): added. This function remove latest `n' variables from vtable. * iseq.c: catch up iseq data changes. * proc.c: ditto. * class.c (keyword_error): export as rb_keyword_error(). * common.mk: depend vm_args.c for vm.o. * hash.c (rb_hash_has_key): export. * internal.h: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@48239 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2014-11-02 21:02:55 +03:00
/* required argument. do nothing */
COMPILE_ERROR(ERROR_ARGS "unreachable");
goto ng;
* rewrite method/block parameter fitting logic to optimize keyword arguments/parameters and a splat argument. [Feature #10440] (Details are described in this ticket) Most of complex part is moved to vm_args.c. Now, ISeq#to_a does not catch up new instruction format. * vm_core.h: change iseq data structures. * introduce rb_call_info_kw_arg_t to represent keyword arguments. * add rb_call_info_t::kw_arg. * rename rb_iseq_t::arg_post_len to rb_iseq_t::arg_post_num. * rename rb_iseq_t::arg_keywords to arg_keyword_num. * rename rb_iseq_t::arg_keyword to rb_iseq_t::arg_keyword_bits. to represent keyword bitmap parameter index. This bitmap parameter shows that which keyword parameters are given or not given (0 for given). It is refered by `checkkeyword' instruction described bellow. * rename rb_iseq_t::arg_keyword_check to rb_iseq_t::arg_keyword_rest to represent keyword rest parameter index. * add rb_iseq_t::arg_keyword_default_values to represent default keyword values. * rename VM_CALL_ARGS_SKIP_SETUP to VM_CALL_ARGS_SIMPLE to represent (ci->flag & (SPLAT|BLOCKARG)) && ci->blockiseq == NULL && ci->kw_arg == NULL. * vm_insnhelper.c, vm_args.c: rewrite with refactoring. * rewrite splat argument code. * rewrite keyword arguments/parameters code. * merge method and block parameter fitting code into one code base. * vm.c, vm_eval.c: catch up these changes. * compile.c (new_callinfo): callinfo requires kw_arg parameter. * compile.c (compile_array_): check the last argument Hash object or not. If Hash object and all keys are Symbol literals, they are compiled to keyword arguments. * insns.def (checkkeyword): add new instruction. This instruction check the availability of corresponding keyword. For example, a method "def foo k1: 'v1'; end" is cimpiled to the following instructions. 0000 checkkeyword 2, 0 # check k1 is given. 0003 branchif 9 # if given, jump to address #9 0005 putstring "v1" 0007 setlocal_OP__WC__0 3 # k1 = 'v1' 0009 trace 8 0011 putnil 0012 trace 16 0014 leave * insns.def (opt_send_simple): removed and add new instruction "opt_send_without_block". * parse.y (new_args_tail_gen): reorder variables. Before this patch, a method "def foo(k1: 1, kr1:, k2: 2, **krest, &b)" has parameter variables "k1, kr1, k2, &b, internal_id, krest", but this patch reorders to "kr1, k1, k2, internal_id, krest, &b". (locate a block variable at last) * parse.y (vtable_pop): added. This function remove latest `n' variables from vtable. * iseq.c: catch up iseq data changes. * proc.c: ditto. * class.c (keyword_error): export as rb_keyword_error(). * common.mk: depend vm_args.c for vm.o. * hash.c (rb_hash_has_key): export. * internal.h: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@48239 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2014-11-02 21:02:55 +03:00
}
else if (nd_type(default_value) == NODE_LIT ||
nd_type(default_value) == NODE_NIL ||
nd_type(default_value) == NODE_TRUE ||
nd_type(default_value) == NODE_FALSE) {
COMPILE_ERROR(ERROR_ARGS "unreachable");
goto ng;
* rewrite method/block parameter fitting logic to optimize keyword arguments/parameters and a splat argument. [Feature #10440] (Details are described in this ticket) Most of complex part is moved to vm_args.c. Now, ISeq#to_a does not catch up new instruction format. * vm_core.h: change iseq data structures. * introduce rb_call_info_kw_arg_t to represent keyword arguments. * add rb_call_info_t::kw_arg. * rename rb_iseq_t::arg_post_len to rb_iseq_t::arg_post_num. * rename rb_iseq_t::arg_keywords to arg_keyword_num. * rename rb_iseq_t::arg_keyword to rb_iseq_t::arg_keyword_bits. to represent keyword bitmap parameter index. This bitmap parameter shows that which keyword parameters are given or not given (0 for given). It is refered by `checkkeyword' instruction described bellow. * rename rb_iseq_t::arg_keyword_check to rb_iseq_t::arg_keyword_rest to represent keyword rest parameter index. * add rb_iseq_t::arg_keyword_default_values to represent default keyword values. * rename VM_CALL_ARGS_SKIP_SETUP to VM_CALL_ARGS_SIMPLE to represent (ci->flag & (SPLAT|BLOCKARG)) && ci->blockiseq == NULL && ci->kw_arg == NULL. * vm_insnhelper.c, vm_args.c: rewrite with refactoring. * rewrite splat argument code. * rewrite keyword arguments/parameters code. * merge method and block parameter fitting code into one code base. * vm.c, vm_eval.c: catch up these changes. * compile.c (new_callinfo): callinfo requires kw_arg parameter. * compile.c (compile_array_): check the last argument Hash object or not. If Hash object and all keys are Symbol literals, they are compiled to keyword arguments. * insns.def (checkkeyword): add new instruction. This instruction check the availability of corresponding keyword. For example, a method "def foo k1: 'v1'; end" is cimpiled to the following instructions. 0000 checkkeyword 2, 0 # check k1 is given. 0003 branchif 9 # if given, jump to address #9 0005 putstring "v1" 0007 setlocal_OP__WC__0 3 # k1 = 'v1' 0009 trace 8 0011 putnil 0012 trace 16 0014 leave * insns.def (opt_send_simple): removed and add new instruction "opt_send_without_block". * parse.y (new_args_tail_gen): reorder variables. Before this patch, a method "def foo(k1: 1, kr1:, k2: 2, **krest, &b)" has parameter variables "k1, kr1, k2, &b, internal_id, krest", but this patch reorders to "kr1, k1, k2, internal_id, krest, &b". (locate a block variable at last) * parse.y (vtable_pop): added. This function remove latest `n' variables from vtable. * iseq.c: catch up iseq data changes. * proc.c: ditto. * class.c (keyword_error): export as rb_keyword_error(). * common.mk: depend vm_args.c for vm.o. * hash.c (rb_hash_has_key): export. * internal.h: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@48239 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2014-11-02 21:02:55 +03:00
}
else {
/* if keywordcheck(_kw_bits, nth_keyword)
* kw = default_value
* end
*/
int kw_bits_idx = body->local_table_size - body->param.keyword->bits_start;
int keyword_idx = body->param.keyword->num;
* rewrite method/block parameter fitting logic to optimize keyword arguments/parameters and a splat argument. [Feature #10440] (Details are described in this ticket) Most of complex part is moved to vm_args.c. Now, ISeq#to_a does not catch up new instruction format. * vm_core.h: change iseq data structures. * introduce rb_call_info_kw_arg_t to represent keyword arguments. * add rb_call_info_t::kw_arg. * rename rb_iseq_t::arg_post_len to rb_iseq_t::arg_post_num. * rename rb_iseq_t::arg_keywords to arg_keyword_num. * rename rb_iseq_t::arg_keyword to rb_iseq_t::arg_keyword_bits. to represent keyword bitmap parameter index. This bitmap parameter shows that which keyword parameters are given or not given (0 for given). It is refered by `checkkeyword' instruction described bellow. * rename rb_iseq_t::arg_keyword_check to rb_iseq_t::arg_keyword_rest to represent keyword rest parameter index. * add rb_iseq_t::arg_keyword_default_values to represent default keyword values. * rename VM_CALL_ARGS_SKIP_SETUP to VM_CALL_ARGS_SIMPLE to represent (ci->flag & (SPLAT|BLOCKARG)) && ci->blockiseq == NULL && ci->kw_arg == NULL. * vm_insnhelper.c, vm_args.c: rewrite with refactoring. * rewrite splat argument code. * rewrite keyword arguments/parameters code. * merge method and block parameter fitting code into one code base. * vm.c, vm_eval.c: catch up these changes. * compile.c (new_callinfo): callinfo requires kw_arg parameter. * compile.c (compile_array_): check the last argument Hash object or not. If Hash object and all keys are Symbol literals, they are compiled to keyword arguments. * insns.def (checkkeyword): add new instruction. This instruction check the availability of corresponding keyword. For example, a method "def foo k1: 'v1'; end" is cimpiled to the following instructions. 0000 checkkeyword 2, 0 # check k1 is given. 0003 branchif 9 # if given, jump to address #9 0005 putstring "v1" 0007 setlocal_OP__WC__0 3 # k1 = 'v1' 0009 trace 8 0011 putnil 0012 trace 16 0014 leave * insns.def (opt_send_simple): removed and add new instruction "opt_send_without_block". * parse.y (new_args_tail_gen): reorder variables. Before this patch, a method "def foo(k1: 1, kr1:, k2: 2, **krest, &b)" has parameter variables "k1, kr1, k2, &b, internal_id, krest", but this patch reorders to "kr1, k1, k2, internal_id, krest, &b". (locate a block variable at last) * parse.y (vtable_pop): added. This function remove latest `n' variables from vtable. * iseq.c: catch up iseq data changes. * proc.c: ditto. * class.c (keyword_error): export as rb_keyword_error(). * common.mk: depend vm_args.c for vm.o. * hash.c (rb_hash_has_key): export. * internal.h: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@48239 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2014-11-02 21:02:55 +03:00
ADD_INSN2(ret, line, checkkeyword, INT2FIX(kw_bits_idx + VM_ENV_DATA_SIZE - 1), INT2FIX(keyword_idx));
* rewrite method/block parameter fitting logic to optimize keyword arguments/parameters and a splat argument. [Feature #10440] (Details are described in this ticket) Most of complex part is moved to vm_args.c. Now, ISeq#to_a does not catch up new instruction format. * vm_core.h: change iseq data structures. * introduce rb_call_info_kw_arg_t to represent keyword arguments. * add rb_call_info_t::kw_arg. * rename rb_iseq_t::arg_post_len to rb_iseq_t::arg_post_num. * rename rb_iseq_t::arg_keywords to arg_keyword_num. * rename rb_iseq_t::arg_keyword to rb_iseq_t::arg_keyword_bits. to represent keyword bitmap parameter index. This bitmap parameter shows that which keyword parameters are given or not given (0 for given). It is refered by `checkkeyword' instruction described bellow. * rename rb_iseq_t::arg_keyword_check to rb_iseq_t::arg_keyword_rest to represent keyword rest parameter index. * add rb_iseq_t::arg_keyword_default_values to represent default keyword values. * rename VM_CALL_ARGS_SKIP_SETUP to VM_CALL_ARGS_SIMPLE to represent (ci->flag & (SPLAT|BLOCKARG)) && ci->blockiseq == NULL && ci->kw_arg == NULL. * vm_insnhelper.c, vm_args.c: rewrite with refactoring. * rewrite splat argument code. * rewrite keyword arguments/parameters code. * merge method and block parameter fitting code into one code base. * vm.c, vm_eval.c: catch up these changes. * compile.c (new_callinfo): callinfo requires kw_arg parameter. * compile.c (compile_array_): check the last argument Hash object or not. If Hash object and all keys are Symbol literals, they are compiled to keyword arguments. * insns.def (checkkeyword): add new instruction. This instruction check the availability of corresponding keyword. For example, a method "def foo k1: 'v1'; end" is cimpiled to the following instructions. 0000 checkkeyword 2, 0 # check k1 is given. 0003 branchif 9 # if given, jump to address #9 0005 putstring "v1" 0007 setlocal_OP__WC__0 3 # k1 = 'v1' 0009 trace 8 0011 putnil 0012 trace 16 0014 leave * insns.def (opt_send_simple): removed and add new instruction "opt_send_without_block". * parse.y (new_args_tail_gen): reorder variables. Before this patch, a method "def foo(k1: 1, kr1:, k2: 2, **krest, &b)" has parameter variables "k1, kr1, k2, &b, internal_id, krest", but this patch reorders to "kr1, k1, k2, internal_id, krest, &b". (locate a block variable at last) * parse.y (vtable_pop): added. This function remove latest `n' variables from vtable. * iseq.c: catch up iseq data changes. * proc.c: ditto. * class.c (keyword_error): export as rb_keyword_error(). * common.mk: depend vm_args.c for vm.o. * hash.c (rb_hash_has_key): export. * internal.h: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@48239 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2014-11-02 21:02:55 +03:00
ADD_INSNL(ret, line, branchif, end_label);
CHECK(COMPILE_POPPED(ret, "keyword default argument", node->nd_body));
* rewrite method/block parameter fitting logic to optimize keyword arguments/parameters and a splat argument. [Feature #10440] (Details are described in this ticket) Most of complex part is moved to vm_args.c. Now, ISeq#to_a does not catch up new instruction format. * vm_core.h: change iseq data structures. * introduce rb_call_info_kw_arg_t to represent keyword arguments. * add rb_call_info_t::kw_arg. * rename rb_iseq_t::arg_post_len to rb_iseq_t::arg_post_num. * rename rb_iseq_t::arg_keywords to arg_keyword_num. * rename rb_iseq_t::arg_keyword to rb_iseq_t::arg_keyword_bits. to represent keyword bitmap parameter index. This bitmap parameter shows that which keyword parameters are given or not given (0 for given). It is refered by `checkkeyword' instruction described bellow. * rename rb_iseq_t::arg_keyword_check to rb_iseq_t::arg_keyword_rest to represent keyword rest parameter index. * add rb_iseq_t::arg_keyword_default_values to represent default keyword values. * rename VM_CALL_ARGS_SKIP_SETUP to VM_CALL_ARGS_SIMPLE to represent (ci->flag & (SPLAT|BLOCKARG)) && ci->blockiseq == NULL && ci->kw_arg == NULL. * vm_insnhelper.c, vm_args.c: rewrite with refactoring. * rewrite splat argument code. * rewrite keyword arguments/parameters code. * merge method and block parameter fitting code into one code base. * vm.c, vm_eval.c: catch up these changes. * compile.c (new_callinfo): callinfo requires kw_arg parameter. * compile.c (compile_array_): check the last argument Hash object or not. If Hash object and all keys are Symbol literals, they are compiled to keyword arguments. * insns.def (checkkeyword): add new instruction. This instruction check the availability of corresponding keyword. For example, a method "def foo k1: 'v1'; end" is cimpiled to the following instructions. 0000 checkkeyword 2, 0 # check k1 is given. 0003 branchif 9 # if given, jump to address #9 0005 putstring "v1" 0007 setlocal_OP__WC__0 3 # k1 = 'v1' 0009 trace 8 0011 putnil 0012 trace 16 0014 leave * insns.def (opt_send_simple): removed and add new instruction "opt_send_without_block". * parse.y (new_args_tail_gen): reorder variables. Before this patch, a method "def foo(k1: 1, kr1:, k2: 2, **krest, &b)" has parameter variables "k1, kr1, k2, &b, internal_id, krest", but this patch reorders to "kr1, k1, k2, internal_id, krest, &b". (locate a block variable at last) * parse.y (vtable_pop): added. This function remove latest `n' variables from vtable. * iseq.c: catch up iseq data changes. * proc.c: ditto. * class.c (keyword_error): export as rb_keyword_error(). * common.mk: depend vm_args.c for vm.o. * hash.c (rb_hash_has_key): export. * internal.h: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@48239 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2014-11-02 21:02:55 +03:00
ADD_LABEL(ret, end_label);
}
break;
}
case NODE_DSYM:{
compile_dstr(iseq, ret, node);
if (!popped) {
ADD_INSN(ret, line, intern);
}
else {
ADD_INSN(ret, line, pop);
}
break;
}
case NODE_ATTRASGN:{
DECL_ANCHOR(recv);
DECL_ANCHOR(args);
unsigned int flag = 0;
ID mid = node->nd_mid;
VALUE argc;
LABEL *else_label = NULL;
VALUE branches = Qfalse;
/* optimization shortcut
* obj["literal"] = value -> opt_aset_with(obj, "literal", value)
*/
if (mid == idASET && !private_recv_p(node) && node->nd_args &&
nd_type(node->nd_args) == NODE_LIST && node->nd_args->nd_alen == 2 &&
nd_type(node->nd_args->nd_head) == NODE_STR &&
ISEQ_COMPILE_DATA(iseq)->current_block == NULL &&
!ISEQ_COMPILE_DATA(iseq)->option->frozen_string_literal &&
ISEQ_COMPILE_DATA(iseq)->option->specialized_instruction)
{
VALUE str = rb_fstring(node->nd_args->nd_head->nd_lit);
CHECK(COMPILE(ret, "recv", node->nd_recv));
CHECK(COMPILE(ret, "value", node->nd_args->nd_next->nd_head));
if (!popped) {
ADD_INSN(ret, line, swap);
ADD_INSN1(ret, line, topn, INT2FIX(1));
}
ADD_INSN2(ret, line, opt_aset_with, str,
new_callinfo(iseq, idASET, 2, 0, NULL, FALSE));
RB_OBJ_WRITTEN(iseq, Qundef, str);
ADD_INSN(ret, line, pop);
break;
}
INIT_ANCHOR(recv);
INIT_ANCHOR(args);
* rewrite method/block parameter fitting logic to optimize keyword arguments/parameters and a splat argument. [Feature #10440] (Details are described in this ticket) Most of complex part is moved to vm_args.c. Now, ISeq#to_a does not catch up new instruction format. * vm_core.h: change iseq data structures. * introduce rb_call_info_kw_arg_t to represent keyword arguments. * add rb_call_info_t::kw_arg. * rename rb_iseq_t::arg_post_len to rb_iseq_t::arg_post_num. * rename rb_iseq_t::arg_keywords to arg_keyword_num. * rename rb_iseq_t::arg_keyword to rb_iseq_t::arg_keyword_bits. to represent keyword bitmap parameter index. This bitmap parameter shows that which keyword parameters are given or not given (0 for given). It is refered by `checkkeyword' instruction described bellow. * rename rb_iseq_t::arg_keyword_check to rb_iseq_t::arg_keyword_rest to represent keyword rest parameter index. * add rb_iseq_t::arg_keyword_default_values to represent default keyword values. * rename VM_CALL_ARGS_SKIP_SETUP to VM_CALL_ARGS_SIMPLE to represent (ci->flag & (SPLAT|BLOCKARG)) && ci->blockiseq == NULL && ci->kw_arg == NULL. * vm_insnhelper.c, vm_args.c: rewrite with refactoring. * rewrite splat argument code. * rewrite keyword arguments/parameters code. * merge method and block parameter fitting code into one code base. * vm.c, vm_eval.c: catch up these changes. * compile.c (new_callinfo): callinfo requires kw_arg parameter. * compile.c (compile_array_): check the last argument Hash object or not. If Hash object and all keys are Symbol literals, they are compiled to keyword arguments. * insns.def (checkkeyword): add new instruction. This instruction check the availability of corresponding keyword. For example, a method "def foo k1: 'v1'; end" is cimpiled to the following instructions. 0000 checkkeyword 2, 0 # check k1 is given. 0003 branchif 9 # if given, jump to address #9 0005 putstring "v1" 0007 setlocal_OP__WC__0 3 # k1 = 'v1' 0009 trace 8 0011 putnil 0012 trace 16 0014 leave * insns.def (opt_send_simple): removed and add new instruction "opt_send_without_block". * parse.y (new_args_tail_gen): reorder variables. Before this patch, a method "def foo(k1: 1, kr1:, k2: 2, **krest, &b)" has parameter variables "k1, kr1, k2, &b, internal_id, krest", but this patch reorders to "kr1, k1, k2, internal_id, krest, &b". (locate a block variable at last) * parse.y (vtable_pop): added. This function remove latest `n' variables from vtable. * iseq.c: catch up iseq data changes. * proc.c: ditto. * class.c (keyword_error): export as rb_keyword_error(). * common.mk: depend vm_args.c for vm.o. * hash.c (rb_hash_has_key): export. * internal.h: ditto. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@48239 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2014-11-02 21:02:55 +03:00
argc = setup_args(iseq, args, node->nd_args, &flag, NULL);
CHECK(!NIL_P(argc));
2019-09-19 17:44:37 +03:00
int asgnflag = COMPILE_RECV(recv, "recv", node);
CHECK(asgnflag != -1);
flag |= (unsigned int)asgnflag;
debugp_param("argc", argc);
debugp_param("nd_mid", ID2SYM(mid));
if (!rb_is_attrset_id(mid)) {
/* safe nav attr */
mid = rb_id_attrset(mid);
else_label = qcall_branch_start(iseq, recv, &branches, node, line);
}
if (!popped) {
ADD_INSN(ret, line, putnil);
ADD_SEQ(ret, recv);
ADD_SEQ(ret, args);
if (flag & VM_CALL_ARGS_BLOCKARG) {
ADD_INSN1(ret, line, topn, INT2FIX(1));
if (flag & VM_CALL_ARGS_SPLAT) {
ADD_INSN1(ret, line, putobject, INT2FIX(-1));
ADD_SEND_WITH_FLAG(ret, line, idAREF, INT2FIX(1), INT2FIX(asgnflag));
}
ADD_INSN1(ret, line, setn, FIXNUM_INC(argc, 3));
ADD_INSN (ret, line, pop);
}
else if (flag & VM_CALL_ARGS_SPLAT) {
ADD_INSN(ret, line, dup);
ADD_INSN1(ret, line, putobject, INT2FIX(-1));
ADD_SEND_WITH_FLAG(ret, line, idAREF, INT2FIX(1), INT2FIX(asgnflag));
ADD_INSN1(ret, line, setn, FIXNUM_INC(argc, 2));
ADD_INSN (ret, line, pop);
}
else {
ADD_INSN1(ret, line, setn, FIXNUM_INC(argc, 1));
}
}
else {
ADD_SEQ(ret, recv);
ADD_SEQ(ret, args);
}
ADD_SEND_WITH_FLAG(ret, line, mid, argc, INT2FIX(flag));
qcall_branch_end(iseq, ret, else_label, branches, node, line);
ADD_INSN(ret, line, pop);
break;
}
case NODE_LAMBDA:{
/* compile same as lambda{...} */
2015-07-22 01:52:59 +03:00
const rb_iseq_t *block = NEW_CHILD_ISEQ(node->nd_body, make_name_for_block(iseq), ISEQ_TYPE_BLOCK, line);
VALUE argc = INT2FIX(0);
2015-07-22 01:52:59 +03:00
ADD_INSN1(ret, line, putspecialobject, INT2FIX(VM_SPECIAL_OBJECT_VMCORE));
ADD_CALL_WITH_BLOCK(ret, line, idLambda, argc, block);
RB_OBJ_WRITTEN(iseq, Qundef, (VALUE)block);
if (popped) {
ADD_INSN(ret, line, pop);
}
break;
}
case NODE_METHREF:
CHECK(COMPILE(ret, "receiver", node->nd_recv));
ADD_ELEM(ret, &new_insn_body(iseq, line, BIN(methodref), 1, ID2SYM(node->nd_mid))->link);
if (popped) {
ADD_INSN(ret, line, pop);
}
break;
default:
UNKNOWN_NODE("iseq_compile_each", node, COMPILE_NG);
ng:
debug_node_end();
return COMPILE_NG;
}
debug_node_end();
return COMPILE_OK;
}
/***************************/
/* instruction information */
/***************************/
static int
insn_data_length(INSN *iobj)
{
return insn_len(iobj->insn_id);
}
static int
calc_sp_depth(int depth, INSN *insn)
{
return comptime_insn_stack_increase(depth, insn->insn_id, insn->operands);
}
static VALUE
opobj_inspect(VALUE obj)
{
struct RBasic *r = (struct RBasic *) obj;
if (!SPECIAL_CONST_P(r) && r->klass == 0) {
switch (BUILTIN_TYPE(r)) {
case T_STRING:
obj = rb_str_new_cstr(RSTRING_PTR(obj));
break;
case T_ARRAY:
obj = rb_ary_dup(obj);
break;
}
}
return rb_inspect(obj);
}
static VALUE
insn_data_to_s_detail(INSN *iobj)
{
VALUE str = rb_sprintf("%-20s ", insn_name(iobj->insn_id));
if (iobj->operands) {
const char *types = insn_op_types(iobj->insn_id);
int j;
for (j = 0; types[j]; j++) {
char type = types[j];
switch (type) {
case TS_OFFSET: /* label(destination position) */
{
LABEL *lobj = (LABEL *)OPERAND_AT(iobj, j);
rb_str_catf(str, LABEL_FORMAT, lobj->label_no);
break;
}
break;
case TS_ISEQ: /* iseq */
{
rb_iseq_t *iseq = (rb_iseq_t *)OPERAND_AT(iobj, j);
VALUE val = Qnil;
if (0 && iseq) { /* TODO: invalidate now */
2015-07-22 01:52:59 +03:00
val = (VALUE)iseq;
}
rb_str_concat(str, opobj_inspect(val));
}
break;
case TS_LINDEX:
case TS_NUM: /* ulong */
case TS_VALUE: /* VALUE */
{
VALUE v = OPERAND_AT(iobj, j);
rb_str_concat(str, opobj_inspect(v));
break;
}
case TS_ID: /* ID */
rb_str_concat(str, opobj_inspect(OPERAND_AT(iobj, j)));
break;
case TS_GENTRY:
{
struct rb_global_entry *entry = (struct rb_global_entry *)
(OPERAND_AT(iobj, j) & (~1));
rb_str_append(str, rb_id2str(entry->id));
break;
}
case TS_IC: /* inline cache */
case TS_ISE: /* inline storage entry */
rb_str_catf(str, "<ic:%d>", FIX2INT(OPERAND_AT(iobj, j)));
break;
case TS_CALLDATA: /* we store these as call infos at compile time */
{
const struct rb_call_info *ci = (struct rb_call_info *)OPERAND_AT(iobj, j);
rb_str_cat2(str, "<calldata:");
if (ci->mid) rb_str_catf(str, "%"PRIsVALUE, rb_id2str(ci->mid));
rb_str_catf(str, ", %d>", ci->orig_argc);
break;
}
case TS_CDHASH: /* case/when condition cache */
rb_str_cat2(str, "<ch>");
break;
case TS_FUNCPTR:
{
void *func = (void *)OPERAND_AT(iobj, j);
#ifdef HAVE_DLADDR
Dl_info info;
if (dladdr(func, &info) && info.dli_sname) {
rb_str_cat2(str, info.dli_sname);
break;
}
#endif
rb_str_catf(str, "<%p>", func);
}
break;
default:{
rb_raise(rb_eSyntaxError, "unknown operand type: %c", type);
}
}
if (types[j + 1]) {
rb_str_cat2(str, ", ");
}
}
}
return str;
}
static void
dump_disasm_list(const LINK_ELEMENT *link)
{
dump_disasm_list_with_cursor(link, NULL, NULL);
}
static void
dump_disasm_list_with_cursor(const LINK_ELEMENT *link, const LINK_ELEMENT *curr, const LABEL *dest)
{
int pos = 0;
INSN *iobj;
LABEL *lobj;
VALUE str;
printf("-- raw disasm--------\n");
while (link) {
if (curr) printf(curr == link ? "*" : " ");
switch (link->type) {
case ISEQ_ELEMENT_INSN:
{
iobj = (INSN *)link;
str = insn_data_to_s_detail(iobj);
printf("%04d %-65s(%4u)\n", pos, StringValueCStr(str), iobj->insn_info.line_no);
pos += insn_data_length(iobj);
break;
}
case ISEQ_ELEMENT_LABEL:
{
lobj = (LABEL *)link;
printf(LABEL_FORMAT" [sp: %d]%s\n", lobj->label_no, lobj->sp,
dest == lobj ? " <---" : "");
break;
}
case ISEQ_ELEMENT_TRACE:
{
TRACE *trace = (TRACE *)link;
printf("trace: %0x\n", trace->event);
break;
}
case ISEQ_ELEMENT_ADJUST:
{
ADJUST *adjust = (ADJUST *)link;
printf("adjust: [label: %d]\n", adjust->label ? adjust->label->label_no : -1);
break;
}
default:
/* ignore */
rb_raise(rb_eSyntaxError, "dump_disasm_list error: %ld\n", FIX2LONG(link->type));
}
link = link->next;
}
printf("---------------------\n");
fflush(stdout);
}
* probes.d: add DTrace probe declarations. [ruby-core:27448] * array.c (empty_ary_alloc, ary_new): added array create DTrace probe. * compile.c (rb_insns_name): allowing DTrace probes to access instruction sequence name. * Makefile.in: translate probes.d file to appropriate header file. * common.mk: declare dependencies on the DTrace header. * configure.in: add a test for existence of DTrace. * eval.c (setup_exception): add a probe for when an exception is raised. * gc.c: Add DTrace probes for mark begin and end, and sweep begin and end. * hash.c (empty_hash_alloc): Add a probe for hash allocation. * insns.def: Add probes for function entry and return. * internal.h: function declaration for compile.c change. * load.c (rb_f_load): add probes for `load` entry and exit, require entry and exit, and wrapping search_required for load path search. * object.c (rb_obj_alloc): added a probe for general object creation. * parse.y (yycompile0): added a probe around parse and compile phase. * string.c (empty_str_alloc, str_new): DTrace probes for string allocation. * test/dtrace/*: tests for DTrace probes. * vm.c (vm_invoke_proc): add probes for function return on exception raise, hash create, and instruction sequence execution. * vm_core.h: add probe declarations for function entry and exit. * vm_dump.c: add probes header file. * vm_eval.c (vm_call0_cfunc, vm_call0_cfunc_with_frame): add probe on function entry and return. * vm_exec.c: expose instruction number to instruction name function. * vm_insnshelper.c: add function entry and exit probes for cfunc methods. * vm_insnhelper.h: vm usage information is always collected, so uncomment the functions. 12 19:14:50 2012 Akinori MUSHA <knu@iDaemons.org> * configure.in (isinf, isnan): isinf() and isnan() are macros on DragonFly which cannot be found by AC_REPLACE_FUNCS(). This workaround enforces the fact that they exist on DragonFly. 12 15:59:38 2012 Shugo Maeda <shugo@ruby-lang.org> * vm_core.h (rb_call_info_t::refinements), compile.c (new_callinfo), vm_insnhelper.c (vm_search_method): revert r37616 because it's too slow. [ruby-dev:46477] * test/ruby/test_refinement.rb (test_inline_method_cache): skip the test until the bug is fixed efficiently. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@37631 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2012-11-13 01:52:12 +04:00
const char *
rb_insns_name(int i)
{
return insn_name(i);
* probes.d: add DTrace probe declarations. [ruby-core:27448] * array.c (empty_ary_alloc, ary_new): added array create DTrace probe. * compile.c (rb_insns_name): allowing DTrace probes to access instruction sequence name. * Makefile.in: translate probes.d file to appropriate header file. * common.mk: declare dependencies on the DTrace header. * configure.in: add a test for existence of DTrace. * eval.c (setup_exception): add a probe for when an exception is raised. * gc.c: Add DTrace probes for mark begin and end, and sweep begin and end. * hash.c (empty_hash_alloc): Add a probe for hash allocation. * insns.def: Add probes for function entry and return. * internal.h: function declaration for compile.c change. * load.c (rb_f_load): add probes for `load` entry and exit, require entry and exit, and wrapping search_required for load path search. * object.c (rb_obj_alloc): added a probe for general object creation. * parse.y (yycompile0): added a probe around parse and compile phase. * string.c (empty_str_alloc, str_new): DTrace probes for string allocation. * test/dtrace/*: tests for DTrace probes. * vm.c (vm_invoke_proc): add probes for function return on exception raise, hash create, and instruction sequence execution. * vm_core.h: add probe declarations for function entry and exit. * vm_dump.c: add probes header file. * vm_eval.c (vm_call0_cfunc, vm_call0_cfunc_with_frame): add probe on function entry and return. * vm_exec.c: expose instruction number to instruction name function. * vm_insnshelper.c: add function entry and exit probes for cfunc methods. * vm_insnhelper.h: vm usage information is always collected, so uncomment the functions. 12 19:14:50 2012 Akinori MUSHA <knu@iDaemons.org> * configure.in (isinf, isnan): isinf() and isnan() are macros on DragonFly which cannot be found by AC_REPLACE_FUNCS(). This workaround enforces the fact that they exist on DragonFly. 12 15:59:38 2012 Shugo Maeda <shugo@ruby-lang.org> * vm_core.h (rb_call_info_t::refinements), compile.c (new_callinfo), vm_insnhelper.c (vm_search_method): revert r37616 because it's too slow. [ruby-dev:46477] * test/ruby/test_refinement.rb (test_inline_method_cache): skip the test until the bug is fixed efficiently. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@37631 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2012-11-13 01:52:12 +04:00
}
VALUE
rb_insns_name_array(void)
{
VALUE ary = rb_ary_new_capa(VM_INSTRUCTION_SIZE);
int i;
for (i = 0; i < VM_INSTRUCTION_SIZE; i++) {
rb_ary_push(ary, rb_fstring_cstr(insn_name(i)));
}
return rb_obj_freeze(ary);
}
static LABEL *
register_label(rb_iseq_t *iseq, struct st_table *labels_table, VALUE obj)
{
LABEL *label = 0;
st_data_t tmp;
obj = rb_to_symbol_type(obj);
if (st_lookup(labels_table, obj, &tmp) == 0) {
label = NEW_LABEL(0);
st_insert(labels_table, obj, (st_data_t)label);
}
else {
label = (LABEL *)tmp;
}
LABEL_REF(label);
return label;
}
static VALUE
get_exception_sym2type(VALUE sym)
{
#undef rb_intern
#define rb_intern(str) rb_intern_const(str)
static VALUE symRescue, symEnsure, symRetry;
static VALUE symBreak, symRedo, symNext;
if (symRescue == 0) {
symRescue = ID2SYM(rb_intern("rescue"));
symEnsure = ID2SYM(rb_intern("ensure"));
symRetry = ID2SYM(rb_intern("retry"));
symBreak = ID2SYM(rb_intern("break"));
symRedo = ID2SYM(rb_intern("redo"));
symNext = ID2SYM(rb_intern("next"));
}
if (sym == symRescue) return CATCH_TYPE_RESCUE;
if (sym == symEnsure) return CATCH_TYPE_ENSURE;
if (sym == symRetry) return CATCH_TYPE_RETRY;
if (sym == symBreak) return CATCH_TYPE_BREAK;
if (sym == symRedo) return CATCH_TYPE_REDO;
if (sym == symNext) return CATCH_TYPE_NEXT;
rb_raise(rb_eSyntaxError, "invalid exception symbol: %+"PRIsVALUE, sym);
return 0;
}
static int
iseq_build_from_ary_exception(rb_iseq_t *iseq, struct st_table *labels_table,
VALUE exception)
{
int i;
for (i=0; i<RARRAY_LEN(exception); i++) {
2015-07-22 01:52:59 +03:00
const rb_iseq_t *eiseq;
VALUE v, type;
LABEL *lstart, *lend, *lcont;
unsigned int sp;
v = rb_to_array_type(RARRAY_AREF(exception, i));
if (RARRAY_LEN(v) != 6) {
rb_raise(rb_eSyntaxError, "wrong exception entry");
}
type = get_exception_sym2type(RARRAY_AREF(v, 0));
if (RARRAY_AREF(v, 1) == Qnil) {
2015-07-22 01:52:59 +03:00
eiseq = NULL;
}
else {
eiseq = rb_iseqw_to_iseq(rb_iseq_load(RARRAY_AREF(v, 1), (VALUE)iseq, Qnil));
}
lstart = register_label(iseq, labels_table, RARRAY_AREF(v, 2));
lend = register_label(iseq, labels_table, RARRAY_AREF(v, 3));
lcont = register_label(iseq, labels_table, RARRAY_AREF(v, 4));
sp = NUM2UINT(RARRAY_AREF(v, 5));
/* TODO: Dirty Hack! Fix me */
if (type == CATCH_TYPE_RESCUE ||
type == CATCH_TYPE_BREAK ||
type == CATCH_TYPE_NEXT) {
++sp;
}
lcont->sp = sp;
2015-07-22 01:52:59 +03:00
ADD_CATCH_ENTRY(type, lstart, lend, eiseq, lcont);
RB_GC_GUARD(v);
}
return COMPILE_OK;
}
static struct st_table *
insn_make_insn_table(void)
{
struct st_table *table;
int i;
table = st_init_numtable_with_size(VM_INSTRUCTION_SIZE);
for (i=0; i<VM_INSTRUCTION_SIZE; i++) {
st_insert(table, ID2SYM(rb_intern(insn_name(i))), i);
}
return table;
}
2015-07-22 01:52:59 +03:00
static const rb_iseq_t *
iseq_build_load_iseq(const rb_iseq_t *iseq, VALUE op)
{
2015-07-22 01:52:59 +03:00
VALUE iseqw;
const rb_iseq_t *loaded_iseq;
if (RB_TYPE_P(op, T_ARRAY)) {
2015-07-22 01:52:59 +03:00
iseqw = rb_iseq_load(op, (VALUE)iseq, Qnil);
}
else if (CLASS_OF(op) == rb_cISeq) {
2015-07-22 01:52:59 +03:00
iseqw = op;
}
else {
rb_raise(rb_eSyntaxError, "ISEQ is required");
}
2015-07-22 01:52:59 +03:00
loaded_iseq = rb_iseqw_to_iseq(iseqw);
return loaded_iseq;
}
static VALUE
iseq_build_callinfo_from_hash(rb_iseq_t *iseq, VALUE op)
{
ID mid = 0;
int orig_argc = 0;
unsigned int flag = 0;
struct rb_call_info_kw_arg *kw_arg = 0;
if (!NIL_P(op)) {
VALUE vmid = rb_hash_aref(op, ID2SYM(rb_intern("mid")));
VALUE vflag = rb_hash_aref(op, ID2SYM(rb_intern("flag")));
VALUE vorig_argc = rb_hash_aref(op, ID2SYM(rb_intern("orig_argc")));
VALUE vkw_arg = rb_hash_aref(op, ID2SYM(rb_intern("kw_arg")));
if (!NIL_P(vmid)) mid = SYM2ID(vmid);
if (!NIL_P(vflag)) flag = NUM2UINT(vflag);
if (!NIL_P(vorig_argc)) orig_argc = FIX2INT(vorig_argc);
if (!NIL_P(vkw_arg)) {
int i;
int len = RARRAY_LENINT(vkw_arg);
size_t n = rb_call_info_kw_arg_bytes(len);
kw_arg = xmalloc(n);
kw_arg->keyword_len = len;
for (i = 0; i < len; i++) {
VALUE kw = RARRAY_AREF(vkw_arg, i);
SYM2ID(kw); /* make immortal */
kw_arg->keywords[i] = kw;
}
}
}
return (VALUE)new_callinfo(iseq, mid, orig_argc, flag, kw_arg, (flag & VM_CALL_ARGS_SIMPLE) == 0);
}
static rb_event_flag_t
event_name_to_flag(VALUE sym)
{
#define CHECK_EVENT(ev) if (sym == ID2SYM(rb_intern(#ev))) return ev;
CHECK_EVENT(RUBY_EVENT_LINE);
CHECK_EVENT(RUBY_EVENT_CLASS);
CHECK_EVENT(RUBY_EVENT_END);
CHECK_EVENT(RUBY_EVENT_CALL);
CHECK_EVENT(RUBY_EVENT_RETURN);
CHECK_EVENT(RUBY_EVENT_B_CALL);
CHECK_EVENT(RUBY_EVENT_B_RETURN);
#undef CHECK_EVENT
return RUBY_EVENT_NONE;
}
static int
iseq_build_from_ary_body(rb_iseq_t *iseq, LINK_ANCHOR *const anchor,
VALUE body, VALUE labels_wrapper)
{
/* TODO: body should be frozen */
long i, len = RARRAY_LEN(body);
struct st_table *labels_table = DATA_PTR(labels_wrapper);
int j;
int line_no = 0;
int ret = COMPILE_OK;
/*
* index -> LABEL *label
*/
static struct st_table *insn_table;
if (insn_table == 0) {
insn_table = insn_make_insn_table();
}
for (i=0; i<len; i++) {
VALUE obj = RARRAY_AREF(body, i);
if (SYMBOL_P(obj)) {
rb_event_flag_t event;
if ((event = event_name_to_flag(obj)) != RUBY_EVENT_NONE) {
ADD_TRACE(anchor, event);
}
else {
LABEL *label = register_label(iseq, labels_table, obj);
ADD_LABEL(anchor, label);
}
}
else if (FIXNUM_P(obj)) {
line_no = NUM2INT(obj);
}
else if (RB_TYPE_P(obj, T_ARRAY)) {
VALUE *argv = 0;
int argc = RARRAY_LENINT(obj) - 1;
st_data_t insn_id;
VALUE insn;
insn = (argc < 0) ? Qnil : RARRAY_AREF(obj, 0);
if (st_lookup(insn_table, (st_data_t)insn, &insn_id) == 0) {
/* TODO: exception */
COMPILE_ERROR(iseq, line_no,
"unknown instruction: %+"PRIsVALUE, insn);
ret = COMPILE_NG;
break;
}
if (argc != insn_len((VALUE)insn_id)-1) {
COMPILE_ERROR(iseq, line_no,
"operand size mismatch");
ret = COMPILE_NG;
break;
}
if (argc > 0) {
argv = compile_data_alloc2(iseq, sizeof(VALUE), argc);
for (j=0; j<argc; j++) {
VALUE op = rb_ary_entry(obj, j+1);
switch (insn_op_type((VALUE)insn_id, j)) {
case TS_OFFSET: {
LABEL *label = register_label(iseq, labels_table, op);
argv[j] = (VALUE)label;
break;
}
case TS_LINDEX:
case TS_NUM:
(void)NUM2INT(op);
argv[j] = op;
break;
case TS_VALUE:
argv[j] = op;
RB_OBJ_WRITTEN(iseq, Qundef, op);
break;
case TS_ISEQ:
{
if (op != Qnil) {
VALUE v = (VALUE)iseq_build_load_iseq(iseq, op);
argv[j] = v;
RB_OBJ_WRITTEN(iseq, Qundef, v);
}
else {
argv[j] = 0;
}
}
break;
case TS_GENTRY:
op = rb_to_symbol_type(op);
argv[j] = (VALUE)rb_global_entry(SYM2ID(op));
break;
case TS_ISE:
FL_SET(iseq, ISEQ_MARKABLE_ISEQ);
2019-07-16 01:19:52 +03:00
/* fall through */
case TS_IC:
argv[j] = op;
if (NUM2UINT(op) >= iseq->body->is_size) {
2015-07-22 01:52:59 +03:00
iseq->body->is_size = NUM2INT(op) + 1;
}
break;
case TS_CALLDATA:
argv[j] = iseq_build_callinfo_from_hash(iseq, op);
break;
case TS_ID:
argv[j] = rb_to_symbol_type(op);
break;
case TS_CDHASH:
{
int i;
VALUE map = rb_hash_new_with_size(RARRAY_LEN(op)/2);
RHASH_TBL_RAW(map)->type = &cdhash_type;
op = rb_to_array_type(op);
for (i=0; i<RARRAY_LEN(op); i+=2) {
VALUE key = RARRAY_AREF(op, i);
VALUE sym = RARRAY_AREF(op, i+1);
LABEL *label =
register_label(iseq, labels_table, sym);
rb_hash_aset(map, key, (VALUE)label | 1);
}
RB_GC_GUARD(op);
argv[j] = map;
RB_OBJ_WRITTEN(iseq, Qundef, map);
}
break;
case TS_FUNCPTR:
{
#if SIZEOF_VALUE <= SIZEOF_LONG
long funcptr = NUM2LONG(op);
#else
LONG_LONG funcptr = NUM2LL(op);
#endif
argv[j] = (VALUE)funcptr;
}
break;
default:
rb_raise(rb_eSyntaxError, "unknown operand: %c", insn_op_type((VALUE)insn_id, j));
}
}
}
ADD_ELEM(anchor,
(LINK_ELEMENT*)new_insn_core(iseq, line_no,
(enum ruby_vminsn_type)insn_id, argc, argv));
}
else {
rb_raise(rb_eTypeError, "unexpected object for instruction");
}
}
DATA_PTR(labels_wrapper) = 0;
validate_labels(iseq, labels_table);
if (!ret) return ret;
return iseq_setup(iseq, anchor);
}
#define CHECK_ARRAY(v) rb_to_array_type(v)
#define CHECK_SYMBOL(v) rb_to_symbol_type(v)
static int
int_param(int *dst, VALUE param, VALUE sym)
{
VALUE val = rb_hash_aref(param, sym);
if (FIXNUM_P(val)) {
*dst = FIX2INT(val);
return TRUE;
}
else if (!NIL_P(val)) {
rb_raise(rb_eTypeError, "invalid %+"PRIsVALUE" Fixnum: %+"PRIsVALUE,
sym, val);
}
return FALSE;
}
static const struct rb_iseq_param_keyword *
iseq_build_kw(rb_iseq_t *iseq, VALUE params, VALUE keywords)
{
int i, j;
int len = RARRAY_LENINT(keywords);
int default_len;
VALUE key, sym, default_val;
VALUE *dvs;
ID *ids;
struct rb_iseq_param_keyword *keyword = ZALLOC(struct rb_iseq_param_keyword);
2015-07-22 01:52:59 +03:00
iseq->body->param.flags.has_kw = TRUE;
keyword->num = len;
#define SYM(s) ID2SYM(rb_intern(#s))
(void)int_param(&keyword->bits_start, params, SYM(kwbits));
i = keyword->bits_start - keyword->num;
ids = (ID *)&iseq->body->local_table[i];
#undef SYM
/* required args */
for (i = 0; i < len; i++) {
VALUE val = RARRAY_AREF(keywords, i);
if (!SYMBOL_P(val)) {
goto default_values;
}
ids[i] = SYM2ID(val);
keyword->required_num++;
}
default_values: /* note: we intentionally preserve `i' from previous loop */
default_len = len - i;
if (default_len == 0) {
keyword->table = ids;
return keyword;
}
else if (default_len < 0) {
UNREACHABLE;
}
dvs = ALLOC_N(VALUE, (unsigned int)default_len);
for (j = 0; i < len; i++, j++) {
key = RARRAY_AREF(keywords, i);
CHECK_ARRAY(key);
switch (RARRAY_LEN(key)) {
case 1:
sym = RARRAY_AREF(key, 0);
default_val = Qundef;
break;
case 2:
sym = RARRAY_AREF(key, 0);
default_val = RARRAY_AREF(key, 1);
break;
default:
rb_raise(rb_eTypeError, "keyword default has unsupported len %+"PRIsVALUE, key);
}
ids[i] = SYM2ID(sym);
dvs[j] = default_val;
}
keyword->table = ids;
keyword->default_values = dvs;
return keyword;
}
2015-07-22 01:52:59 +03:00
void
rb_iseq_mark_insn_storage(struct iseq_compile_data_storage *storage)
{
INSN *iobj = 0;
size_t size = sizeof(INSN);
unsigned int pos = 0;
while (storage) {
#ifdef STRICT_ALIGNMENT
size_t padding = calc_padding((void *)&storage->buff[pos], size);
#else
const size_t padding = 0; /* expected to be optimized by compiler */
#endif /* STRICT_ALIGNMENT */
size_t offset = pos + size + padding;
if (offset > storage->size || offset > storage->pos) {
pos = 0;
storage = storage->next;
2019-09-27 04:20:56 +03:00
}
else {
#ifdef STRICT_ALIGNMENT
pos += (int)padding;
#endif /* STRICT_ALIGNMENT */
iobj = (INSN *)&storage->buff[pos];
if (iobj->operands) {
int j;
const char *types = insn_op_types(iobj->insn_id);
2019-09-27 04:20:56 +03:00
for (j = 0; types[j]; j++) {
char type = types[j];
2019-09-27 04:20:56 +03:00
switch (type) {
case TS_CDHASH:
case TS_ISEQ:
case TS_VALUE:
{
VALUE op = OPERAND_AT(iobj, j);
if (!SPECIAL_CONST_P(op)) {
rb_gc_mark(op);
}
break;
}
2019-09-27 04:20:56 +03:00
default:
break;
}
}
}
pos += (int)size;
}
}
}
void
rb_iseq_build_from_ary(rb_iseq_t *iseq, VALUE misc, VALUE locals, VALUE params,
VALUE exception, VALUE body)
{
#define SYM(s) ID2SYM(rb_intern(#s))
int i, len;
unsigned int arg_size, local_size, stack_max;
ID *tbl;
struct st_table *labels_table = st_init_numtable();
VALUE labels_wrapper = Data_Wrap_Struct(0, 0, st_free_table, labels_table);
VALUE arg_opt_labels = rb_hash_aref(params, SYM(opt));
VALUE keywords = rb_hash_aref(params, SYM(keyword));
VALUE sym_arg_rest = ID2SYM(rb_intern("#arg_rest"));
DECL_ANCHOR(anchor);
INIT_ANCHOR(anchor);
len = RARRAY_LENINT(locals);
2015-07-22 01:52:59 +03:00
iseq->body->local_table_size = len;
iseq->body->local_table = tbl = len > 0 ? (ID *)ALLOC_N(ID, iseq->body->local_table_size) : NULL;
for (i = 0; i < len; i++) {
VALUE lv = RARRAY_AREF(locals, i);
if (sym_arg_rest == lv) {
tbl[i] = 0;
}
else {
tbl[i] = FIXNUM_P(lv) ? (ID)FIX2LONG(lv) : SYM2ID(CHECK_SYMBOL(lv));
}
}
2015-07-22 01:52:59 +03:00
#define INT_PARAM(F) int_param(&iseq->body->param.F, params, SYM(F))
if (INT_PARAM(lead_num)) {
2015-07-22 01:52:59 +03:00
iseq->body->param.flags.has_lead = TRUE;
}
2015-07-22 01:52:59 +03:00
if (INT_PARAM(post_num)) iseq->body->param.flags.has_post = TRUE;
if (INT_PARAM(post_start)) iseq->body->param.flags.has_post = TRUE;
if (INT_PARAM(rest_start)) iseq->body->param.flags.has_rest = TRUE;
if (INT_PARAM(block_start)) iseq->body->param.flags.has_block = TRUE;
#undef INT_PARAM
{
#define INT_PARAM(F) F = (int_param(&x, misc, SYM(F)) ? (unsigned int)x : 0)
int x;
INT_PARAM(arg_size);
INT_PARAM(local_size);
INT_PARAM(stack_max);
#undef INT_PARAM
}
if (RB_TYPE_P(arg_opt_labels, T_ARRAY)) {
len = RARRAY_LENINT(arg_opt_labels);
2015-07-22 01:52:59 +03:00
iseq->body->param.flags.has_opt = !!(len - 1 >= 0);
2015-07-22 01:52:59 +03:00
if (iseq->body->param.flags.has_opt) {
VALUE *opt_table = ALLOC_N(VALUE, len);
for (i = 0; i < len; i++) {
VALUE ent = RARRAY_AREF(arg_opt_labels, i);
LABEL *label = register_label(iseq, labels_table, ent);
opt_table[i] = (VALUE)label;
}
iseq->body->param.opt_num = len - 1;
iseq->body->param.opt_table = opt_table;
}
}
else if (!NIL_P(arg_opt_labels)) {
rb_raise(rb_eTypeError, ":opt param is not an array: %+"PRIsVALUE,
arg_opt_labels);
}
if (RB_TYPE_P(keywords, T_ARRAY)) {
iseq->body->param.keyword = iseq_build_kw(iseq, params, keywords);
}
else if (!NIL_P(keywords)) {
rb_raise(rb_eTypeError, ":keywords param is not an array: %+"PRIsVALUE,
keywords);
}
if (Qtrue == rb_hash_aref(params, SYM(ambiguous_param0))) {
2015-07-22 01:52:59 +03:00
iseq->body->param.flags.ambiguous_param0 = TRUE;
}
if (int_param(&i, params, SYM(kwrest))) {
struct rb_iseq_param_keyword *keyword = (struct rb_iseq_param_keyword *)iseq->body->param.keyword;
if (keyword == NULL) {
iseq->body->param.keyword = keyword = ZALLOC(struct rb_iseq_param_keyword);
}
keyword->rest_start = i;
iseq->body->param.flags.has_kwrest = TRUE;
}
#undef SYM
iseq_calc_param_size(iseq);
/* exception */
iseq_build_from_ary_exception(iseq, labels_table, exception);
/* body */
iseq_build_from_ary_body(iseq, anchor, body, labels_wrapper);
iseq->body->param.size = arg_size;
iseq->body->local_table_size = local_size;
iseq->body->stack_max = stack_max;
}
/* for parser */
int
rb_dvar_defined(ID id, const rb_iseq_t *iseq)
{
if (iseq) {
const struct rb_iseq_constant_body *body = iseq->body;
while (body->type == ISEQ_TYPE_BLOCK ||
body->type == ISEQ_TYPE_RESCUE ||
body->type == ISEQ_TYPE_ENSURE ||
body->type == ISEQ_TYPE_EVAL ||
body->type == ISEQ_TYPE_MAIN
) {
unsigned int i;
for (i = 0; i < body->local_table_size; i++) {
if (body->local_table[i] == id) {
return 1;
}
}
iseq = body->parent_iseq;
body = iseq->body;
}
}
return 0;
}
int
rb_local_defined(ID id, const rb_iseq_t *iseq)
{
if (iseq) {
unsigned int i;
const struct rb_iseq_constant_body *const body = iseq->body->local_iseq->body;
for (i=0; i<body->local_table_size; i++) {
if (body->local_table[i] == id) {
return 1;
}
}
}
return 0;
}
static int
caller_location(VALUE *path, VALUE *realpath)
{
const rb_execution_context_t *ec = GET_EC();
const rb_control_frame_t *const cfp =
rb_vm_get_ruby_level_next_cfp(ec, ec->cfp);
if (cfp) {
int line = rb_vm_get_sourceline(cfp);
*path = rb_iseq_path(cfp->iseq);
*realpath = rb_iseq_realpath(cfp->iseq);
return line;
}
else {
*path = rb_fstring_lit("<compiled>");
*realpath = *path;
return 1;
}
}
typedef struct {
VALUE arg;
rb_insn_func_t func;
int line;
} accessor_args;
2015-07-22 01:52:59 +03:00
static const rb_iseq_t *
method_for_self(VALUE name, VALUE arg, rb_insn_func_t func,
void (*build)(rb_iseq_t *, LINK_ANCHOR *, const void *))
{
VALUE path, realpath;
accessor_args acc;
acc.arg = arg;
acc.func = func;
acc.line = caller_location(&path, &realpath);
struct rb_iseq_new_with_callback_callback_func *ifunc =
rb_iseq_new_with_callback_new_callback(build, &acc);
return rb_iseq_new_with_callback(ifunc,
rb_sym2str(name), path, realpath,
INT2FIX(acc.line), 0, ISEQ_TYPE_METHOD, 0);
}
static void
for_self_aref(rb_iseq_t *iseq, LINK_ANCHOR *ret, const void *a)
{
const accessor_args *const args = (void *)a;
const int line = args->line;
struct rb_iseq_constant_body *const body = iseq->body;
iseq_set_local_table(iseq, 0);
body->param.lead_num = 0;
body->param.size = 0;
ADD_INSN1(ret, line, putobject, args->arg);
ADD_INSN1(ret, line, opt_call_c_function, (VALUE)args->func);
}
static void
for_self_aset(rb_iseq_t *iseq, LINK_ANCHOR *ret, const void *a)
{
const accessor_args *const args = (void *)a;
const int line = args->line;
struct rb_iseq_constant_body *const body = iseq->body;
static const ID vars[] = {1, idUScore};
iseq_set_local_table(iseq, vars);
body->param.lead_num = 1;
body->param.size = 1;
ADD_GETLOCAL(ret, line, numberof(vars)-1, 0);
ADD_INSN1(ret, line, putobject, args->arg);
ADD_INSN1(ret, line, opt_call_c_function, (VALUE)args->func);
ADD_INSN(ret, line, pop);
}
/*
* func (index) -> (value)
*/
2015-07-22 01:52:59 +03:00
const rb_iseq_t *
rb_method_for_self_aref(VALUE name, VALUE arg, rb_insn_func_t func)
{
return method_for_self(name, arg, func, for_self_aref);
}
/*
* func (index, value) -> (index, value)
*/
2015-07-22 01:52:59 +03:00
const rb_iseq_t *
rb_method_for_self_aset(VALUE name, VALUE arg, rb_insn_func_t func)
{
return method_for_self(name, arg, func, for_self_aset);
}
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
/* ISeq binary format */
#ifndef IBF_ISEQ_DEBUG
#define IBF_ISEQ_DEBUG 0
#endif
#ifndef IBF_ISEQ_ENABLE_LOCAL_BUFFER
#define IBF_ISEQ_ENABLE_LOCAL_BUFFER 0
#endif
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
typedef unsigned int ibf_offset_t;
#define IBF_OFFSET(ptr) ((ibf_offset_t)(VALUE)(ptr))
#define IBF_MAJOR_VERSION ISEQ_MAJOR_VERSION
#if RUBY_DEVEL
#define IBF_DEVEL_VERSION 2
#define IBF_MINOR_VERSION (ISEQ_MINOR_VERSION * 10000 + IBF_DEVEL_VERSION)
#else
#define IBF_MINOR_VERSION ISEQ_MINOR_VERSION
#endif
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
struct ibf_header {
char magic[4]; /* YARB */
unsigned int major_version;
unsigned int minor_version;
unsigned int size;
unsigned int extra_size;
unsigned int iseq_list_size;
unsigned int global_object_list_size;
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
ibf_offset_t iseq_list_offset;
ibf_offset_t global_object_list_offset;
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
};
struct ibf_dump_buffer {
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
VALUE str;
VALUE obj_list; /* [objs] */
};
struct ibf_dump {
VALUE iseq_list; /* [iseqs] */
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
st_table *iseq_table; /* iseq -> iseq number */
struct ibf_dump_buffer global_buffer;
struct ibf_dump_buffer *current_buffer;
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
};
rb_iseq_t * iseq_alloc(void);
struct ibf_load_buffer {
const char *buff;
ibf_offset_t size;
VALUE obj_list; /* [obj0, ...] */
unsigned int obj_list_size;
ibf_offset_t obj_list_offset;
};
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
struct ibf_load {
const struct ibf_header *header;
VALUE iseq_list; /* [iseq0, ...] */
struct ibf_load_buffer global_buffer;
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
VALUE loader_obj;
rb_iseq_t *iseq;
VALUE str;
struct ibf_load_buffer *current_buffer;
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
};
static ibf_offset_t
ibf_dump_pos(struct ibf_dump *dump)
{
long pos = RSTRING_LEN(dump->current_buffer->str);
#if SIZEOF_LONG > SIZEOF_INT
if (pos >= UINT_MAX) {
rb_raise(rb_eRuntimeError, "dump size exceeds");
}
#endif
return (unsigned int)pos;
}
static void
ibf_dump_align(struct ibf_dump *dump, size_t align)
{
ibf_offset_t pos = ibf_dump_pos(dump);
if (pos % align) {
static const char padding[sizeof(VALUE)];
size_t size = align - ((size_t)pos % align);
#if SIZEOF_LONG > SIZEOF_INT
if (pos + size >= UINT_MAX) {
rb_raise(rb_eRuntimeError, "dump size exceeds");
}
#endif
for (; size > sizeof(padding); size -= sizeof(padding)) {
rb_str_cat(dump->current_buffer->str, padding, sizeof(padding));
}
rb_str_cat(dump->current_buffer->str, padding, size);
}
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
static ibf_offset_t
ibf_dump_write(struct ibf_dump *dump, const void *buff, unsigned long size)
{
ibf_offset_t pos = ibf_dump_pos(dump);
rb_str_cat(dump->current_buffer->str, (const char *)buff, size);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
/* TODO: overflow check */
return pos;
}
static ibf_offset_t
ibf_dump_write_byte(struct ibf_dump *dump, unsigned char byte)
{
return ibf_dump_write(dump, &byte, sizeof(unsigned char));
}
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
static void
ibf_dump_overwrite(struct ibf_dump *dump, void *buff, unsigned int size, long offset)
{
VALUE str = dump->current_buffer->str;
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
char *ptr = RSTRING_PTR(str);
if ((unsigned long)(size + offset) > (unsigned long)RSTRING_LEN(str))
rb_bug("ibf_dump_overwrite: overflow");
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
memcpy(ptr + offset, buff, size);
}
static void *
ibf_load_alloc(const struct ibf_load *load, ibf_offset_t offset, size_t x, size_t y)
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
{
void *buff = ruby_xmalloc2(x, y);
size_t size = x * y;
memcpy(buff, load->current_buffer->buff + offset, size);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
return buff;
}
#define IBF_W_ALIGN(type) (RUBY_ALIGNOF(type) > 1 ? ibf_dump_align(dump, RUBY_ALIGNOF(type)) : (void)0)
#define IBF_W(b, type, n) (IBF_W_ALIGN(type), (type *)(VALUE)IBF_WP(b, type, n))
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
#define IBF_WV(variable) ibf_dump_write(dump, &(variable), sizeof(variable))
#define IBF_WP(b, type, n) ibf_dump_write(dump, (b), sizeof(type) * (n))
#define IBF_R(val, type, n) (type *)ibf_load_alloc(load, IBF_OFFSET(val), sizeof(type), (n))
#define IBF_ZERO(variable) memset(&(variable), 0, sizeof(variable))
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
static int
ibf_table_lookup(struct st_table *table, st_data_t key)
{
st_data_t val;
if (st_lookup(table, key, &val)) {
return (int)val;
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
else {
return -1;
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
}
static int
ibf_table_index(struct st_table *table, st_data_t key)
{
int index = ibf_table_lookup(table, key);
if (index < 0) { /* not found */
index = (int)table->num_entries;
st_insert(table, key, (st_data_t)index);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
return index;
}
/* dump/load generic */
static void ibf_dump_object_list(struct ibf_dump *dump, ibf_offset_t *obj_list_offset, unsigned int *obj_list_size);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
static VALUE ibf_load_object(const struct ibf_load *load, VALUE object_index);
static rb_iseq_t *ibf_load_iseq(const struct ibf_load *load, const rb_iseq_t *index_iseq);
static VALUE
ibf_dump_object_list_new(void)
{
VALUE obj_list = rb_ary_tmp_new(1);
rb_ary_push(obj_list, Qnil); /* 0th is nil */
return obj_list;
}
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
static VALUE
ibf_dump_object(struct ibf_dump *dump, VALUE obj)
{
VALUE obj_list = dump->current_buffer->obj_list;
long index = RARRAY_LEN(obj_list);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
long i;
for (i=0; i<index; i++) {
if (RARRAY_AREF(obj_list, i) == obj) return (VALUE)i; /* dedup */
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
rb_ary_push(obj_list, obj);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
return (VALUE)index;
}
static VALUE
ibf_dump_id(struct ibf_dump *dump, ID id)
{
if (id == 0 || rb_id2name(id) == NULL) {
return 0;
}
return ibf_dump_object(dump, rb_id2sym(id));
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
static ID
ibf_load_id(const struct ibf_load *load, const ID id_index)
{
if (id_index == 0) {
return 0;
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
VALUE sym = ibf_load_object(load, id_index);
return rb_sym2id(sym);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
/* dump/load: code */
static VALUE
ibf_dump_calldata(struct ibf_dump *dump, const struct rb_call_data *cd)
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
{
return (cd->ci.flag & VM_CALL_KWARG) ? Qtrue : Qfalse;
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
static ibf_offset_t ibf_dump_iseq_each(struct ibf_dump *dump, const rb_iseq_t *iseq);
static int
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
ibf_dump_iseq(struct ibf_dump *dump, const rb_iseq_t *iseq)
{
if (iseq == NULL) {
return -1;
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
else {
int iseq_index = ibf_table_lookup(dump->iseq_table, (st_data_t)iseq);
if (iseq_index < 0) {
iseq_index = ibf_table_index(dump->iseq_table, (st_data_t)iseq);
rb_ary_push(dump->iseq_list, (VALUE)iseq);
}
return iseq_index;
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
}
static VALUE
ibf_dump_gentry(struct ibf_dump *dump, const struct rb_global_entry *entry)
{
return (VALUE)ibf_dump_id(dump, entry->id);
}
static VALUE
ibf_load_gentry(const struct ibf_load *load, const struct rb_global_entry *entry)
{
ID gid = ibf_load_id(load, (ID)(VALUE)entry);
return (VALUE)rb_global_entry(gid);
}
static unsigned char
ibf_load_byte(const struct ibf_load *load, ibf_offset_t *offset)
{
if (*offset >= load->current_buffer->size) { rb_raise(rb_eRuntimeError, "invalid bytecode"); }
return (unsigned char)load->current_buffer->buff[(*offset)++];
}
/*
* Small uint serialization
* 0x00000000_00000000 - 0x00000000_0000007f: 1byte | XXXX XXX1 |
* 0x00000000_00000080 - 0x00000000_00003fff: 2byte | XXXX XX10 | XXXX XXXX |
* 0x00000000_00004000 - 0x00000000_001fffff: 3byte | XXXX X100 | XXXX XXXX | XXXX XXXX |
* 0x00000000_00020000 - 0x00000000_0fffffff: 4byte | XXXX 1000 | XXXX XXXX | XXXX XXXX | XXXX XXXX |
* ...
* 0x00010000_00000000 - 0x00ffffff_ffffffff: 8byte | 1000 0000 | XXXX XXXX | XXXX XXXX | XXXX XXXX | XXXX XXXX | XXXX XXXX | XXXX XXXX | XXXX XXXX |
* 0x01000000_00000000 - 0xffffffff_ffffffff: 9byte | 0000 0000 | XXXX XXXX | XXXX XXXX | XXXX XXXX | XXXX XXXX | XXXX XXXX | XXXX XXXX | XXXX XXXX | XXXX XXXX |
*/
static void
ibf_dump_write_small_value(struct ibf_dump *dump, VALUE x)
{
if (sizeof(VALUE) > 8 || CHAR_BIT != 8) {
ibf_dump_write(dump, &x, sizeof(VALUE));
}
enum { max_byte_length = sizeof(VALUE) + 1 };
unsigned char bytes[max_byte_length];
ibf_offset_t n;
for (n = 0; n < sizeof(VALUE) && (x >> (7 - n)); n++, x >>= 8) {
bytes[max_byte_length - 1 - n] = (unsigned char)x;
}
x <<= 1;
x |= 1;
x <<= n;
bytes[max_byte_length - 1 - n] = (unsigned char)x;
n++;
ibf_dump_write(dump, bytes + max_byte_length - n, n);
}
static VALUE
ibf_load_small_value(const struct ibf_load *load, ibf_offset_t *offset)
{
if (sizeof(VALUE) > 8 || CHAR_BIT != 8) {
union { char s[sizeof(VALUE)]; VALUE v; } x;
memcpy(x.s, load->current_buffer->buff + *offset, sizeof(VALUE));
*offset += sizeof(VALUE);
return x.v;
}
enum { max_byte_length = sizeof(VALUE) + 1 };
const unsigned char *buffer = (const unsigned char *)load->current_buffer->buff;
const unsigned char c = buffer[*offset];
ibf_offset_t n =
c & 1 ? 1 :
c == 0 ? 9 : ntz_int32(c) + 1;
VALUE x = (VALUE)c >> n;
if (*offset + n > load->current_buffer->size) {
rb_raise(rb_eRuntimeError, "invalid byte sequence");
}
ibf_offset_t i;
for (i = 1; i < n; i++) {
x <<= 8;
x |= (VALUE)buffer[*offset + i];
}
*offset += n;
return x;
}
static ibf_offset_t
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
ibf_dump_code(struct ibf_dump *dump, const rb_iseq_t *iseq)
{
const struct rb_iseq_constant_body *const body = iseq->body;
const int iseq_size = body->iseq_size;
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
int code_index;
const VALUE *orig_code = rb_iseq_original_iseq(iseq);
ibf_offset_t offset = ibf_dump_pos(dump);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
for (code_index=0; code_index<iseq_size;) {
const VALUE insn = orig_code[code_index++];
const char *types = insn_op_types(insn);
int op_index;
/* opcode */
if (insn >= 0x100) { rb_raise(rb_eRuntimeError, "invalid instruction"); }
ibf_dump_write_small_value(dump, insn);
/* operands */
for (op_index=0; types[op_index]; op_index++, code_index++) {
VALUE op = orig_code[code_index];
switch (types[op_index]) {
case TS_CDHASH:
case TS_VALUE:
ibf_dump_write_small_value(dump, ibf_dump_object(dump, op));
break;
case TS_ISEQ:
{
VALUE index = (VALUE)ibf_dump_iseq(dump, (const rb_iseq_t *)op);
ibf_dump_write_small_value(dump, index);
}
break;
case TS_IC:
case TS_ISE:
{
unsigned int i;
for (i=0; i<body->is_size; i++) {
if (op == (VALUE)&body->is_entries[i]) {
break;
}
}
ibf_dump_write_small_value(dump, (VALUE)i);
}
break;
case TS_CALLDATA:
{
VALUE callinfo = ibf_dump_calldata(dump, (const struct rb_call_data *)op);
/* ibf_dump_calldata() always returns either Qtrue or Qfalse */
ibf_dump_write_byte(dump, callinfo == Qtrue);
}
break;
case TS_ID:
ibf_dump_write_small_value(dump, ibf_dump_id(dump, (ID)op));
break;
case TS_GENTRY:
ibf_dump_write_small_value(dump, ibf_dump_gentry(dump, (const struct rb_global_entry *)op));
break;
case TS_FUNCPTR:
rb_raise(rb_eRuntimeError, "TS_FUNCPTR is not supported");
break;
default:
ibf_dump_write_small_value(dump, op);
break;
}
}
assert(insn_len(insn) == op_index+1);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
return offset;
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
static VALUE *
ibf_load_code(const struct ibf_load *load, const rb_iseq_t *iseq, ibf_offset_t bytecode_offset, ibf_offset_t bytecode_size, unsigned int iseq_size)
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
{
unsigned int code_index;
ibf_offset_t reading_pos = bytecode_offset;
VALUE *code = ALLOC_N(VALUE, iseq_size);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
struct rb_iseq_constant_body *load_body = iseq->body;
struct rb_call_data *cd_entries = load_body->call_data;
struct rb_kwarg_call_data *cd_kw_entries = (struct rb_kwarg_call_data *)&load_body->call_data[load_body->ci_size];
union iseq_inline_storage_entry *is_entries = load_body->is_entries;
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
for (code_index=0; code_index<iseq_size;) {
/* opcode */
const VALUE insn = code[code_index++] = ibf_load_small_value(load, &reading_pos);
const char *types = insn_op_types(insn);
int op_index;
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
/* operands */
for (op_index=0; types[op_index]; op_index++, code_index++) {
switch (types[op_index]) {
case TS_CDHASH:
case TS_VALUE:
{
VALUE op = ibf_load_small_value(load, &reading_pos);
VALUE v = ibf_load_object(load, op);
code[code_index] = v;
if (!SPECIAL_CONST_P(v)) {
RB_OBJ_WRITTEN(iseq, Qundef, v);
FL_SET(iseq, ISEQ_MARKABLE_ISEQ);
}
break;
}
case TS_ISEQ:
{
VALUE op = (VALUE)ibf_load_small_value(load, &reading_pos);
VALUE v = (VALUE)ibf_load_iseq(load, (const rb_iseq_t *)op);
code[code_index] = v;
if (!SPECIAL_CONST_P(v)) {
RB_OBJ_WRITTEN(iseq, Qundef, v);
FL_SET(iseq, ISEQ_MARKABLE_ISEQ);
}
break;
}
case TS_ISE:
FL_SET(iseq, ISEQ_MARKABLE_ISEQ);
2019-07-14 18:30:35 +03:00
/* fall through */
case TS_IC:
{
VALUE op = ibf_load_small_value(load, &reading_pos);
code[code_index] = (VALUE)&is_entries[op];
}
break;
case TS_CALLDATA:
{
unsigned char op = ibf_load_byte(load, &reading_pos);
code[code_index] = op ? (VALUE)cd_kw_entries++ : (VALUE)cd_entries++; /* op is 1 (kw) or 0 (!kw) */
}
break;
case TS_ID:
{
VALUE op = ibf_load_small_value(load, &reading_pos);
code[code_index] = ibf_load_id(load, (ID)(VALUE)op);
}
break;
case TS_GENTRY:
{
VALUE op = ibf_load_small_value(load, &reading_pos);
code[code_index] = ibf_load_gentry(load, (const struct rb_global_entry *)(VALUE)op);
}
break;
case TS_FUNCPTR:
rb_raise(rb_eRuntimeError, "TS_FUNCPTR is not supported");
break;
default:
code[code_index] = ibf_load_small_value(load, &reading_pos);
continue;
}
}
if (insn_len(insn) != op_index+1) {
rb_raise(rb_eRuntimeError, "operand size mismatch");
}
}
load_body->iseq_encoded = code;
load_body->iseq_size = code_index;
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
assert(code_index == iseq_size);
assert(reading_pos == bytecode_offset + bytecode_size);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
return code;
}
static ibf_offset_t
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
ibf_dump_param_opt_table(struct ibf_dump *dump, const rb_iseq_t *iseq)
{
int opt_num = iseq->body->param.opt_num;
if (opt_num > 0) {
IBF_W_ALIGN(VALUE);
return ibf_dump_write(dump, iseq->body->param.opt_table, sizeof(VALUE) * (opt_num + 1));
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
else {
return ibf_dump_pos(dump);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
}
static VALUE *
ibf_load_param_opt_table(const struct ibf_load *load, ibf_offset_t opt_table_offset, int opt_num)
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
{
if (opt_num > 0) {
VALUE *table = ALLOC_N(VALUE, opt_num+1);
MEMCPY(table, load->current_buffer->buff + opt_table_offset, VALUE, opt_num+1);
return table;
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
else {
return NULL;
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
}
static ibf_offset_t
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
ibf_dump_param_keyword(struct ibf_dump *dump, const rb_iseq_t *iseq)
{
const struct rb_iseq_param_keyword *kw = iseq->body->param.keyword;
if (kw) {
struct rb_iseq_param_keyword dump_kw = *kw;
int dv_num = kw->num - kw->required_num;
ID *ids = kw->num > 0 ? ALLOCA_N(ID, kw->num) : NULL;
VALUE *dvs = dv_num > 0 ? ALLOCA_N(VALUE, dv_num) : NULL;
int i;
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
for (i=0; i<kw->num; i++) ids[i] = (ID)ibf_dump_id(dump, kw->table[i]);
for (i=0; i<dv_num; i++) dvs[i] = (VALUE)ibf_dump_object(dump, kw->default_values[i]);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
dump_kw.table = IBF_W(ids, ID, kw->num);
dump_kw.default_values = IBF_W(dvs, VALUE, dv_num);
IBF_W_ALIGN(struct rb_iseq_param_keyword);
return ibf_dump_write(dump, &dump_kw, sizeof(struct rb_iseq_param_keyword) * 1);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
else {
return 0;
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
}
static const struct rb_iseq_param_keyword *
ibf_load_param_keyword(const struct ibf_load *load, ibf_offset_t param_keyword_offset)
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
{
if (param_keyword_offset) {
struct rb_iseq_param_keyword *kw = IBF_R(param_keyword_offset, struct rb_iseq_param_keyword, 1);
ID *ids = IBF_R(kw->table, ID, kw->num);
int dv_num = kw->num - kw->required_num;
VALUE *dvs = IBF_R(kw->default_values, VALUE, dv_num);
int i;
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
for (i=0; i<kw->num; i++) {
ids[i] = ibf_load_id(load, ids[i]);
}
for (i=0; i<dv_num; i++) {
dvs[i] = ibf_load_object(load, dvs[i]);
}
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
kw->table = ids;
kw->default_values = dvs;
return kw;
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
else {
return NULL;
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
}
static ibf_offset_t
ibf_dump_insns_info_body(struct ibf_dump *dump, const rb_iseq_t *iseq)
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
{
ibf_offset_t offset = ibf_dump_pos(dump);
const struct iseq_insn_info_entry *entries = iseq->body->insns_info.body;
unsigned int i;
for (i = 0; i < iseq->body->insns_info.size; i++) {
ibf_dump_write_small_value(dump, entries[i].line_no);
ibf_dump_write_small_value(dump, entries[i].events);
}
return offset;
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
static struct iseq_insn_info_entry *
ibf_load_insns_info_body(const struct ibf_load *load, ibf_offset_t body_offset, unsigned int size)
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
{
ibf_offset_t reading_pos = body_offset;
struct iseq_insn_info_entry *entries = ALLOC_N(struct iseq_insn_info_entry, size);
unsigned int i;
for (i = 0; i < size; i++) {
entries[i].line_no = (int)ibf_load_small_value(load, &reading_pos);
entries[i].events = (rb_event_flag_t)ibf_load_small_value(load, &reading_pos);
}
return entries;
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
static ibf_offset_t
ibf_dump_insns_info_positions(struct ibf_dump *dump, const unsigned int *positions, unsigned int size)
{
ibf_offset_t offset = ibf_dump_pos(dump);
unsigned int last = 0;
unsigned int i;
for (i = 0; i < size; i++) {
ibf_dump_write_small_value(dump, positions[i] - last);
last = positions[i];
}
return offset;
}
static unsigned int *
ibf_load_insns_info_positions(const struct ibf_load *load, ibf_offset_t positions_offset, unsigned int size)
{
ibf_offset_t reading_pos = positions_offset;
unsigned int *positions = ALLOC_N(unsigned int, size);
unsigned int last = 0;
unsigned int i;
for (i = 0; i < size; i++) {
positions[i] = last + (unsigned int)ibf_load_small_value(load, &reading_pos);
last = positions[i];
}
return positions;
}
static ibf_offset_t
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
ibf_dump_local_table(struct ibf_dump *dump, const rb_iseq_t *iseq)
{
const struct rb_iseq_constant_body *const body = iseq->body;
const int size = body->local_table_size;
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
ID *table = ALLOCA_N(ID, size);
int i;
for (i=0; i<size; i++) {
table[i] = ibf_dump_id(dump, body->local_table[i]);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
IBF_W_ALIGN(ID);
return ibf_dump_write(dump, table, sizeof(ID) * size);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
static ID *
ibf_load_local_table(const struct ibf_load *load, ibf_offset_t local_table_offset, int size)
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
{
if (size > 0) {
ID *table = IBF_R(local_table_offset, ID, size);
int i;
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
for (i=0; i<size; i++) {
table[i] = ibf_load_id(load, table[i]);
}
return table;
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
else {
return NULL;
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
}
static ibf_offset_t
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
ibf_dump_catch_table(struct ibf_dump *dump, const rb_iseq_t *iseq)
{
const struct iseq_catch_table *table = iseq->body->catch_table;
if (table) {
int *iseq_indices = ALLOCA_N(int, table->size);
unsigned int i;
for (i=0; i<table->size; i++) {
iseq_indices[i] = ibf_dump_iseq(dump, table->entries[i].iseq);
}
const ibf_offset_t offset = ibf_dump_pos(dump);
for (i=0; i<table->size; i++) {
ibf_dump_write_small_value(dump, iseq_indices[i]);
ibf_dump_write_small_value(dump, table->entries[i].type);
ibf_dump_write_small_value(dump, table->entries[i].start);
ibf_dump_write_small_value(dump, table->entries[i].end);
ibf_dump_write_small_value(dump, table->entries[i].cont);
ibf_dump_write_small_value(dump, table->entries[i].sp);
}
return offset;
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
else {
return ibf_dump_pos(dump);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
}
static struct iseq_catch_table *
ibf_load_catch_table(const struct ibf_load *load, ibf_offset_t catch_table_offset, unsigned int size)
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
{
if (size) {
struct iseq_catch_table *table = ruby_xmalloc(iseq_catch_table_bytes(size));
table->size = size;
ibf_offset_t reading_pos = catch_table_offset;
unsigned int i;
for (i=0; i<table->size; i++) {
int iseq_index = (int)ibf_load_small_value(load, &reading_pos);
table->entries[i].type = (enum catch_type)ibf_load_small_value(load, &reading_pos);
table->entries[i].start = (unsigned int)ibf_load_small_value(load, &reading_pos);
table->entries[i].end = (unsigned int)ibf_load_small_value(load, &reading_pos);
table->entries[i].cont = (unsigned int)ibf_load_small_value(load, &reading_pos);
table->entries[i].sp = (unsigned int)ibf_load_small_value(load, &reading_pos);
table->entries[i].iseq = ibf_load_iseq(load, (const rb_iseq_t *)(VALUE)iseq_index);
}
return table;
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
else {
return NULL;
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
}
static ibf_offset_t
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
ibf_dump_ci_entries(struct ibf_dump *dump, const rb_iseq_t *iseq)
{
const struct rb_iseq_constant_body *const body = iseq->body;
const unsigned int ci_size = body->ci_size;
const unsigned int ci_kw_size = body->ci_kw_size;
const struct rb_call_data *calls = body->call_data;
const struct rb_kwarg_call_data *kw_calls = (const struct rb_kwarg_call_data *)&body->call_data[ci_size];
ibf_offset_t offset = ibf_dump_pos(dump);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
unsigned int i;
for (i = 0; i < ci_size; i++) {
VALUE mid = ibf_dump_id(dump, calls[i].ci.mid);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
ibf_dump_write_small_value(dump, mid);
ibf_dump_write_small_value(dump, calls[i].ci.flag);
ibf_dump_write_small_value(dump, calls[i].ci.orig_argc);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
for (i = 0; i < ci_kw_size; i++) {
const struct rb_call_info_kw_arg *kw_arg = kw_calls[i].ci_kw.kw_arg;
VALUE mid = ibf_dump_id(dump, kw_calls[i].ci_kw.ci.mid);
ibf_dump_write_small_value(dump, mid);
ibf_dump_write_small_value(dump, kw_calls[i].ci_kw.ci.flag);
ibf_dump_write_small_value(dump, kw_calls[i].ci_kw.ci.orig_argc);
ibf_dump_write_small_value(dump, kw_arg->keyword_len);
int j;
for (j = 0; j < kw_calls[i].ci_kw.kw_arg->keyword_len; j++) {
VALUE keyword = ibf_dump_object(dump, kw_arg->keywords[j]); /* kw_arg->keywords[n] is Symbol */
ibf_dump_write_small_value(dump, keyword);
}
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
return offset;
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
/* note that we dump out rb_call_info but load back rb_call_data */
static struct rb_call_data *
ibf_load_ci_entries(const struct ibf_load *load,
ibf_offset_t ci_entries_offset,
unsigned int ci_size,
unsigned int ci_kw_size)
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
{
ibf_offset_t reading_pos = ci_entries_offset;
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
unsigned int i;
struct rb_call_data *calls =
rb_xcalloc_mul_add_mul(
sizeof(struct rb_call_data), ci_size,
sizeof(struct rb_kwarg_call_data), ci_kw_size);
struct rb_kwarg_call_data *kw_calls = (struct rb_kwarg_call_data *)&calls[ci_size];
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
for (i = 0; i < ci_size; i++) {
VALUE mid_index = ibf_load_small_value(load, &reading_pos);
calls[i].ci.mid = ibf_load_id(load, mid_index);
calls[i].ci.flag = (unsigned int)ibf_load_small_value(load, &reading_pos);
calls[i].ci.orig_argc = (int)ibf_load_small_value(load, &reading_pos);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
for (i = 0; i < ci_kw_size; i++) {
VALUE mid_index = ibf_load_small_value(load, &reading_pos);
kw_calls[i].ci_kw.ci.mid = ibf_load_id(load, mid_index);
kw_calls[i].ci_kw.ci.flag = (unsigned int)ibf_load_small_value(load, &reading_pos);
kw_calls[i].ci_kw.ci.orig_argc = (int)ibf_load_small_value(load, &reading_pos);
int keyword_len = (int)ibf_load_small_value(load, &reading_pos);
kw_calls[i].ci_kw.kw_arg =
rb_xmalloc_mul_add(keyword_len - 1, sizeof(VALUE), sizeof(struct rb_call_info_kw_arg));
kw_calls[i].ci_kw.kw_arg->keyword_len = keyword_len;
int j;
for (j = 0; j < kw_calls[i].ci_kw.kw_arg->keyword_len; j++) {
VALUE keyword = ibf_load_small_value(load, &reading_pos);
kw_calls[i].ci_kw.kw_arg->keywords[j] = ibf_load_object(load, keyword);
}
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
return calls;
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
static ibf_offset_t
ibf_dump_iseq_each(struct ibf_dump *dump, const rb_iseq_t *iseq)
{
assert(dump->current_buffer == &dump->global_buffer);
unsigned int *positions;
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
const struct rb_iseq_constant_body *body = iseq->body;
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
const VALUE location_pathobj_index = ibf_dump_object(dump, body->location.pathobj); /* TODO: freeze */
const VALUE location_base_label_index = ibf_dump_object(dump, body->location.base_label);
const VALUE location_label_index = ibf_dump_object(dump, body->location.label);
#if IBF_ISEQ_ENABLE_LOCAL_BUFFER
ibf_offset_t iseq_start = ibf_dump_pos(dump);
struct ibf_dump_buffer *saved_buffer = dump->current_buffer;
struct ibf_dump_buffer buffer;
buffer.str = rb_str_new(0, 0);
buffer.obj_list = ibf_dump_object_list_new();
dump->current_buffer = &buffer;
#endif
const ibf_offset_t bytecode_offset = ibf_dump_code(dump, iseq);
const ibf_offset_t bytecode_size = ibf_dump_pos(dump) - bytecode_offset;
const ibf_offset_t param_opt_table_offset = ibf_dump_param_opt_table(dump, iseq);
const ibf_offset_t param_keyword_offset = ibf_dump_param_keyword(dump, iseq);
const ibf_offset_t insns_info_body_offset = ibf_dump_insns_info_body(dump, iseq);
positions = rb_iseq_insns_info_decode_positions(iseq->body);
const ibf_offset_t insns_info_positions_offset = ibf_dump_insns_info_positions(dump, positions, body->insns_info.size);
ruby_xfree(positions);
const ibf_offset_t local_table_offset = ibf_dump_local_table(dump, iseq);
const unsigned int catch_table_size = body->catch_table ? body->catch_table->size : 0;
const ibf_offset_t catch_table_offset = ibf_dump_catch_table(dump, iseq);
const int parent_iseq_index = ibf_dump_iseq(dump, iseq->body->parent_iseq);
const int local_iseq_index = ibf_dump_iseq(dump, iseq->body->local_iseq);
const ibf_offset_t ci_entries_offset = ibf_dump_ci_entries(dump, iseq);
#if IBF_ISEQ_ENABLE_LOCAL_BUFFER
ibf_offset_t local_obj_list_offset;
unsigned int local_obj_list_size;
ibf_dump_object_list(dump, &local_obj_list_offset, &local_obj_list_size);
#endif
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
ibf_offset_t body_offset = ibf_dump_pos(dump);
/* dump the constant body */
unsigned int param_flags =
(body->param.flags.has_lead << 0) |
(body->param.flags.has_opt << 1) |
(body->param.flags.has_rest << 2) |
(body->param.flags.has_post << 3) |
(body->param.flags.has_kw << 4) |
(body->param.flags.has_kwrest << 5) |
(body->param.flags.has_block << 6) |
(body->param.flags.ambiguous_param0 << 7) |
(body->param.flags.accepts_no_kwarg << 8);
#if IBF_ISEQ_ENABLE_LOCAL_BUFFER
# define IBF_BODY_OFFSET(x) (x)
#else
# define IBF_BODY_OFFSET(x) (body_offset - (x))
#endif
ibf_dump_write_small_value(dump, body->type);
ibf_dump_write_small_value(dump, body->iseq_size);
ibf_dump_write_small_value(dump, IBF_BODY_OFFSET(bytecode_offset));
ibf_dump_write_small_value(dump, bytecode_size);
ibf_dump_write_small_value(dump, param_flags);
ibf_dump_write_small_value(dump, body->param.size);
ibf_dump_write_small_value(dump, body->param.lead_num);
ibf_dump_write_small_value(dump, body->param.opt_num);
ibf_dump_write_small_value(dump, body->param.rest_start);
ibf_dump_write_small_value(dump, body->param.post_start);
ibf_dump_write_small_value(dump, body->param.post_num);
ibf_dump_write_small_value(dump, body->param.block_start);
ibf_dump_write_small_value(dump, IBF_BODY_OFFSET(param_opt_table_offset));
ibf_dump_write_small_value(dump, param_keyword_offset);
ibf_dump_write_small_value(dump, location_pathobj_index);
ibf_dump_write_small_value(dump, location_base_label_index);
ibf_dump_write_small_value(dump, location_label_index);
ibf_dump_write_small_value(dump, body->location.first_lineno);
ibf_dump_write_small_value(dump, body->location.node_id);
ibf_dump_write_small_value(dump, body->location.code_location.beg_pos.lineno);
ibf_dump_write_small_value(dump, body->location.code_location.beg_pos.column);
ibf_dump_write_small_value(dump, body->location.code_location.end_pos.lineno);
ibf_dump_write_small_value(dump, body->location.code_location.end_pos.column);
ibf_dump_write_small_value(dump, IBF_BODY_OFFSET(insns_info_body_offset));
ibf_dump_write_small_value(dump, IBF_BODY_OFFSET(insns_info_positions_offset));
ibf_dump_write_small_value(dump, body->insns_info.size);
ibf_dump_write_small_value(dump, IBF_BODY_OFFSET(local_table_offset));
ibf_dump_write_small_value(dump, catch_table_size);
ibf_dump_write_small_value(dump, IBF_BODY_OFFSET(catch_table_offset));
ibf_dump_write_small_value(dump, parent_iseq_index);
ibf_dump_write_small_value(dump, local_iseq_index);
ibf_dump_write_small_value(dump, IBF_BODY_OFFSET(ci_entries_offset));
ibf_dump_write_small_value(dump, body->variable.flip_count);
ibf_dump_write_small_value(dump, body->local_table_size);
ibf_dump_write_small_value(dump, body->is_size);
ibf_dump_write_small_value(dump, body->ci_size);
ibf_dump_write_small_value(dump, body->ci_kw_size);
ibf_dump_write_small_value(dump, body->stack_max);
ibf_dump_write_small_value(dump, body->catch_except_p);
#undef IBF_BODY_OFFSET
#if IBF_ISEQ_ENABLE_LOCAL_BUFFER
ibf_offset_t iseq_length_bytes = ibf_dump_pos(dump);
dump->current_buffer = saved_buffer;
ibf_dump_write(dump, RSTRING_PTR(buffer.str), iseq_length_bytes);
ibf_offset_t offset = ibf_dump_pos(dump);
ibf_dump_write_small_value(dump, iseq_start);
ibf_dump_write_small_value(dump, iseq_length_bytes);
ibf_dump_write_small_value(dump, body_offset);
ibf_dump_write_small_value(dump, local_obj_list_offset);
ibf_dump_write_small_value(dump, local_obj_list_size);
return offset;
#else
return body_offset;
#endif
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
static VALUE
ibf_load_location_str(const struct ibf_load *load, VALUE str_index)
{
VALUE str = ibf_load_object(load, str_index);
if (str != Qnil) {
str = rb_fstring(str);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
return str;
}
static void
ibf_load_iseq_each(struct ibf_load *load, rb_iseq_t *iseq, ibf_offset_t offset)
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
{
struct rb_iseq_constant_body *load_body = iseq->body = ZALLOC(struct rb_iseq_constant_body);
ibf_offset_t reading_pos = offset;
#if IBF_ISEQ_ENABLE_LOCAL_BUFFER
struct ibf_load_buffer *saved_buffer = load->current_buffer;
load->current_buffer = &load->global_buffer;
const ibf_offset_t iseq_start = ibf_load_small_value(load, &reading_pos);
const ibf_offset_t iseq_length_bytes = ibf_load_small_value(load, &reading_pos);
const ibf_offset_t body_offset = ibf_load_small_value(load, &reading_pos);
struct ibf_load_buffer buffer;
buffer.buff = load->global_buffer.buff + iseq_start;
buffer.size = iseq_length_bytes;
buffer.obj_list_offset = ibf_load_small_value(load, &reading_pos);
buffer.obj_list_size = ibf_load_small_value(load, &reading_pos);
buffer.obj_list = rb_ary_tmp_new(buffer.obj_list_size);
rb_ary_resize(buffer.obj_list, buffer.obj_list_size);
load->current_buffer = &buffer;
reading_pos = body_offset;
#endif
#if IBF_ISEQ_ENABLE_LOCAL_BUFFER
# define IBF_BODY_OFFSET(x) (x)
#else
# define IBF_BODY_OFFSET(x) (offset - (x))
#endif
const unsigned int type = (unsigned int)ibf_load_small_value(load, &reading_pos);
const unsigned int iseq_size = (unsigned int)ibf_load_small_value(load, &reading_pos);
const ibf_offset_t bytecode_offset = (ibf_offset_t)IBF_BODY_OFFSET(ibf_load_small_value(load, &reading_pos));
const ibf_offset_t bytecode_size = (ibf_offset_t)ibf_load_small_value(load, &reading_pos);
const unsigned int param_flags = (unsigned int)ibf_load_small_value(load, &reading_pos);
const unsigned int param_size = (unsigned int)ibf_load_small_value(load, &reading_pos);
const int param_lead_num = (int)ibf_load_small_value(load, &reading_pos);
const int param_opt_num = (int)ibf_load_small_value(load, &reading_pos);
const int param_rest_start = (int)ibf_load_small_value(load, &reading_pos);
const int param_post_start = (int)ibf_load_small_value(load, &reading_pos);
const int param_post_num = (int)ibf_load_small_value(load, &reading_pos);
const int param_block_start = (int)ibf_load_small_value(load, &reading_pos);
const ibf_offset_t param_opt_table_offset = (ibf_offset_t)IBF_BODY_OFFSET(ibf_load_small_value(load, &reading_pos));
const ibf_offset_t param_keyword_offset = (ibf_offset_t)ibf_load_small_value(load, &reading_pos);
const VALUE location_pathobj_index = ibf_load_small_value(load, &reading_pos);
const VALUE location_base_label_index = ibf_load_small_value(load, &reading_pos);
const VALUE location_label_index = ibf_load_small_value(load, &reading_pos);
const VALUE location_first_lineno = ibf_load_small_value(load, &reading_pos);
const int location_node_id = (int)ibf_load_small_value(load, &reading_pos);
const int location_code_location_beg_pos_lineno = (int)ibf_load_small_value(load, &reading_pos);
const int location_code_location_beg_pos_column = (int)ibf_load_small_value(load, &reading_pos);
const int location_code_location_end_pos_lineno = (int)ibf_load_small_value(load, &reading_pos);
const int location_code_location_end_pos_column = (int)ibf_load_small_value(load, &reading_pos);
const ibf_offset_t insns_info_body_offset = (ibf_offset_t)IBF_BODY_OFFSET(ibf_load_small_value(load, &reading_pos));
const ibf_offset_t insns_info_positions_offset = (ibf_offset_t)IBF_BODY_OFFSET(ibf_load_small_value(load, &reading_pos));
const unsigned int insns_info_size = (unsigned int)ibf_load_small_value(load, &reading_pos);
const ibf_offset_t local_table_offset = (ibf_offset_t)IBF_BODY_OFFSET(ibf_load_small_value(load, &reading_pos));
const unsigned int catch_table_size = (unsigned int)ibf_load_small_value(load, &reading_pos);
const ibf_offset_t catch_table_offset = (ibf_offset_t)IBF_BODY_OFFSET(ibf_load_small_value(load, &reading_pos));
const int parent_iseq_index = (int)ibf_load_small_value(load, &reading_pos);
const int local_iseq_index = (int)ibf_load_small_value(load, &reading_pos);
const ibf_offset_t ci_entries_offset = (ibf_offset_t)IBF_BODY_OFFSET(ibf_load_small_value(load, &reading_pos));
const rb_snum_t variable_flip_count = (rb_snum_t)ibf_load_small_value(load, &reading_pos);
const unsigned int local_table_size = (unsigned int)ibf_load_small_value(load, &reading_pos);
const unsigned int is_size = (unsigned int)ibf_load_small_value(load, &reading_pos);
const unsigned int ci_size = (unsigned int)ibf_load_small_value(load, &reading_pos);
const unsigned int ci_kw_size = (unsigned int)ibf_load_small_value(load, &reading_pos);
const unsigned int stack_max = (unsigned int)ibf_load_small_value(load, &reading_pos);
const char catch_except_p = (char)ibf_load_small_value(load, &reading_pos);
#undef IBF_BODY_OFFSET
load_body->type = type;
load_body->stack_max = stack_max;
load_body->param.flags.has_lead = (param_flags >> 0) & 1;
load_body->param.flags.has_opt = (param_flags >> 1) & 1;
load_body->param.flags.has_rest = (param_flags >> 2) & 1;
load_body->param.flags.has_post = (param_flags >> 3) & 1;
load_body->param.flags.has_kw = FALSE;
load_body->param.flags.has_kwrest = (param_flags >> 5) & 1;
load_body->param.flags.has_block = (param_flags >> 6) & 1;
load_body->param.flags.ambiguous_param0 = (param_flags >> 7) & 1;
load_body->param.flags.accepts_no_kwarg = (param_flags >> 8) & 1;
load_body->param.size = param_size;
load_body->param.lead_num = param_lead_num;
load_body->param.opt_num = param_opt_num;
load_body->param.rest_start = param_rest_start;
load_body->param.post_start = param_post_start;
load_body->param.post_num = param_post_num;
load_body->param.block_start = param_block_start;
load_body->local_table_size = local_table_size;
load_body->is_size = is_size;
load_body->ci_size = ci_size;
load_body->ci_kw_size = ci_kw_size;
load_body->insns_info.size = insns_info_size;
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
ISEQ_COVERAGE_SET(iseq, Qnil);
ISEQ_ORIGINAL_ISEQ_CLEAR(iseq);
iseq->body->variable.flip_count = variable_flip_count;
load_body->location.first_lineno = location_first_lineno;
load_body->location.node_id = location_node_id;
load_body->location.code_location.beg_pos.lineno = location_code_location_beg_pos_lineno;
load_body->location.code_location.beg_pos.column = location_code_location_beg_pos_column;
load_body->location.code_location.end_pos.lineno = location_code_location_end_pos_lineno;
load_body->location.code_location.end_pos.column = location_code_location_end_pos_column;
load_body->catch_except_p = catch_except_p;
load_body->is_entries = ZALLOC_N(union iseq_inline_storage_entry, is_size);
load_body->call_data = ibf_load_ci_entries(load, ci_entries_offset, ci_size, ci_kw_size);
load_body->param.opt_table = ibf_load_param_opt_table(load, param_opt_table_offset, param_opt_num);
load_body->param.keyword = ibf_load_param_keyword(load, param_keyword_offset);
load_body->param.flags.has_kw = (param_flags >> 4) & 1;
load_body->insns_info.body = ibf_load_insns_info_body(load, insns_info_body_offset, insns_info_size);
load_body->insns_info.positions = ibf_load_insns_info_positions(load, insns_info_positions_offset, insns_info_size);
load_body->local_table = ibf_load_local_table(load, local_table_offset, local_table_size);
load_body->catch_table = ibf_load_catch_table(load, catch_table_offset, catch_table_size);
load_body->parent_iseq = ibf_load_iseq(load, (const rb_iseq_t *)(VALUE)parent_iseq_index);
load_body->local_iseq = ibf_load_iseq(load, (const rb_iseq_t *)(VALUE)local_iseq_index);
ibf_load_code(load, iseq, bytecode_offset, bytecode_size, iseq_size);
#if VM_INSN_INFO_TABLE_IMPL == 2
rb_iseq_insns_info_encode_positions(iseq);
#endif
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
rb_iseq_translate_threaded_code(iseq);
#if IBF_ISEQ_ENABLE_LOCAL_BUFFER
load->current_buffer = &load->global_buffer;
#endif
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
{
VALUE realpath = Qnil, path = ibf_load_object(load, location_pathobj_index);
if (RB_TYPE_P(path, T_STRING)) {
realpath = path = rb_fstring(path);
}
else if (RB_TYPE_P(path, T_ARRAY)) {
VALUE pathobj = path;
if (RARRAY_LEN(pathobj) != 2) {
rb_raise(rb_eRuntimeError, "path object size mismatch");
}
path = rb_fstring(RARRAY_AREF(pathobj, 0));
realpath = RARRAY_AREF(pathobj, 1);
if (!NIL_P(realpath)) {
if (!RB_TYPE_P(realpath, T_STRING)) {
rb_raise(rb_eArgError, "unexpected realpath %"PRIxVALUE
"(%x), path=%+"PRIsVALUE,
realpath, TYPE(realpath), path);
}
realpath = rb_fstring(realpath);
}
}
else {
rb_raise(rb_eRuntimeError, "unexpected path object");
}
rb_iseq_pathobj_set(iseq, path, realpath);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
RB_OBJ_WRITE(iseq, &load_body->location.base_label, ibf_load_location_str(load, location_base_label_index));
RB_OBJ_WRITE(iseq, &load_body->location.label, ibf_load_location_str(load, location_label_index));
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
#if IBF_ISEQ_ENABLE_LOCAL_BUFFER
load->current_buffer = saved_buffer;
#endif
verify_call_cache(iseq);
}
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
static void
ibf_dump_iseq_list(struct ibf_dump *dump, struct ibf_header *header)
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
{
VALUE list = rb_ary_tmp_new(RARRAY_LEN(dump->iseq_list));
long i;
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
for (i = 0; i < RARRAY_LEN(dump->iseq_list); i++) {
ibf_offset_t offset = ibf_dump_iseq_each(dump, (rb_iseq_t *)RARRAY_AREF(dump->iseq_list, i));
rb_ary_push(list, UINT2NUM(offset));
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
long size = RARRAY_LEN(dump->iseq_list);
ibf_offset_t *offsets = ALLOCA_N(ibf_offset_t, size);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
for (i = 0; i < size; i++) {
offsets[i] = NUM2UINT(RARRAY_AREF(list, i));
}
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
ibf_dump_align(dump, sizeof(ibf_offset_t));
header->iseq_list_offset = ibf_dump_write(dump, offsets, sizeof(ibf_offset_t) * size);
header->iseq_list_size = (unsigned int)size;
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
#define IBF_OBJECT_INTERNAL FL_PROMOTED0
/*
* Binary format
* - ibf_object_header
* - ibf_object_xxx (xxx is type)
*/
struct ibf_object_header {
unsigned int type: 5;
unsigned int special_const: 1;
unsigned int frozen: 1;
unsigned int internal: 1;
};
enum ibf_object_class_index {
IBF_OBJECT_CLASS_OBJECT,
IBF_OBJECT_CLASS_ARRAY,
2019-08-08 19:27:52 +03:00
IBF_OBJECT_CLASS_STANDARD_ERROR,
IBF_OBJECT_CLASS_NO_MATCHING_PATTERN_ERROR,
IBF_OBJECT_CLASS_TYPE_ERROR,
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
};
struct ibf_object_regexp {
long srcstr;
char option;
};
struct ibf_object_hash {
long len;
long keyval[FLEX_ARY_LEN];
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
};
struct ibf_object_struct_range {
long class_index;
long len;
long beg;
long end;
int excl;
};
struct ibf_object_bignum {
ssize_t slen;
BDIGIT digits[FLEX_ARY_LEN];
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
};
enum ibf_object_data_type {
2019-08-08 19:27:52 +03:00
IBF_OBJECT_DATA_ENCODING,
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
};
struct ibf_object_complex_rational {
long a, b;
};
struct ibf_object_symbol {
long str;
};
#define IBF_ALIGNED_OFFSET(align, offset) /* offset > 0 */ \
((((offset) - 1) / (align) + 1) * (align))
#define IBF_OBJBODY(type, offset) (const type *)\
ibf_load_check_offset(load, IBF_ALIGNED_OFFSET(RUBY_ALIGNOF(type), offset))
static const void *
ibf_load_check_offset(const struct ibf_load *load, size_t offset)
{
if (offset >= load->current_buffer->size) {
rb_raise(rb_eIndexError, "object offset out of range: %"PRIdSIZE, offset);
}
return load->current_buffer->buff + offset;
}
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
NORETURN(static void ibf_dump_object_unsupported(struct ibf_dump *dump, VALUE obj));
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
static void
ibf_dump_object_unsupported(struct ibf_dump *dump, VALUE obj)
{
char buff[0x100];
rb_raw_obj_info(buff, sizeof(buff), obj);
rb_raise(rb_eNotImpError, "ibf_dump_object_unsupported: %s", buff);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
static VALUE
ibf_load_object_unsupported(const struct ibf_load *load, const struct ibf_object_header *header, ibf_offset_t offset)
{
rb_raise(rb_eArgError, "unsupported");
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
return Qnil;
}
static void
ibf_dump_object_class(struct ibf_dump *dump, VALUE obj)
{
enum ibf_object_class_index cindex;
if (obj == rb_cObject) {
cindex = IBF_OBJECT_CLASS_OBJECT;
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
else if (obj == rb_cArray) {
cindex = IBF_OBJECT_CLASS_ARRAY;
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
else if (obj == rb_eStandardError) {
cindex = IBF_OBJECT_CLASS_STANDARD_ERROR;
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
else if (obj == rb_eNoMatchingPatternError) {
cindex = IBF_OBJECT_CLASS_NO_MATCHING_PATTERN_ERROR;
}
else if (obj == rb_eTypeError) {
cindex = IBF_OBJECT_CLASS_TYPE_ERROR;
}
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
else {
rb_obj_info_dump(obj);
rb_p(obj);
rb_bug("unsupported class");
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
ibf_dump_write_small_value(dump, (VALUE)cindex);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
static VALUE
ibf_load_object_class(const struct ibf_load *load, const struct ibf_object_header *header, ibf_offset_t offset)
{
enum ibf_object_class_index cindex = (enum ibf_object_class_index)ibf_load_small_value(load, &offset);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
switch (cindex) {
case IBF_OBJECT_CLASS_OBJECT:
return rb_cObject;
case IBF_OBJECT_CLASS_ARRAY:
return rb_cArray;
case IBF_OBJECT_CLASS_STANDARD_ERROR:
return rb_eStandardError;
case IBF_OBJECT_CLASS_NO_MATCHING_PATTERN_ERROR:
return rb_eNoMatchingPatternError;
case IBF_OBJECT_CLASS_TYPE_ERROR:
return rb_eTypeError;
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
rb_raise(rb_eArgError, "ibf_load_object_class: unknown class (%d)", (int)cindex);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
static void
ibf_dump_object_float(struct ibf_dump *dump, VALUE obj)
{
double dbl = RFLOAT_VALUE(obj);
(void)IBF_W(&dbl, double, 1);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
static VALUE
ibf_load_object_float(const struct ibf_load *load, const struct ibf_object_header *header, ibf_offset_t offset)
{
const double *dblp = IBF_OBJBODY(double, offset);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
return DBL2NUM(*dblp);
}
static void
ibf_dump_object_string(struct ibf_dump *dump, VALUE obj)
{
long encindex = (long)rb_enc_get_index(obj);
long len = RSTRING_LEN(obj);
const char *ptr = RSTRING_PTR(obj);
if (encindex > RUBY_ENCINDEX_BUILTIN_MAX) {
rb_encoding *enc = rb_enc_from_index((int)encindex);
const char *enc_name = rb_enc_name(enc);
encindex = RUBY_ENCINDEX_BUILTIN_MAX + ibf_dump_object(dump, rb_str_new2(enc_name));
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
ibf_dump_write_small_value(dump, encindex);
ibf_dump_write_small_value(dump, len);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
IBF_WP(ptr, char, len);
}
static VALUE
ibf_load_object_string(const struct ibf_load *load, const struct ibf_object_header *header, ibf_offset_t offset)
{
ibf_offset_t reading_pos = offset;
int encindex = (int)ibf_load_small_value(load, &reading_pos);
const long len = (long)ibf_load_small_value(load, &reading_pos);
const char *ptr = load->current_buffer->buff + reading_pos;
VALUE str = rb_str_new(ptr, len);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
if (encindex > RUBY_ENCINDEX_BUILTIN_MAX) {
VALUE enc_name_str = ibf_load_object(load, encindex - RUBY_ENCINDEX_BUILTIN_MAX);
encindex = rb_enc_find_index(RSTRING_PTR(enc_name_str));
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
rb_enc_associate_index(str, encindex);
if (header->internal) rb_obj_hide(str);
if (header->frozen) str = rb_fstring(str);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
return str;
}
static void
ibf_dump_object_regexp(struct ibf_dump *dump, VALUE obj)
{
VALUE srcstr = RREGEXP_SRC(obj);
struct ibf_object_regexp regexp;
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
regexp.option = (char)rb_reg_options(obj);
regexp.srcstr = (long)ibf_dump_object(dump, srcstr);
ibf_dump_write_byte(dump, (unsigned char)regexp.option);
ibf_dump_write_small_value(dump, regexp.srcstr);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
static VALUE
ibf_load_object_regexp(const struct ibf_load *load, const struct ibf_object_header *header, ibf_offset_t offset)
{
struct ibf_object_regexp regexp;
regexp.option = ibf_load_byte(load, &offset);
regexp.srcstr = ibf_load_small_value(load, &offset);
VALUE srcstr = ibf_load_object(load, regexp.srcstr);
VALUE reg = rb_reg_compile(srcstr, (int)regexp.option, NULL, 0);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
if (header->internal) rb_obj_hide(reg);
if (header->frozen) rb_obj_freeze(reg);
return reg;
}
static void
ibf_dump_object_array(struct ibf_dump *dump, VALUE obj)
{
long i, len = RARRAY_LEN(obj);
ibf_dump_write_small_value(dump, len);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
for (i=0; i<len; i++) {
long index = (long)ibf_dump_object(dump, RARRAY_AREF(obj, i));
ibf_dump_write_small_value(dump, index);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
}
static VALUE
ibf_load_object_array(const struct ibf_load *load, const struct ibf_object_header *header, ibf_offset_t offset)
{
ibf_offset_t reading_pos = offset;
const long len = (long)ibf_load_small_value(load, &reading_pos);
VALUE ary = rb_ary_new_capa(len);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
int i;
for (i=0; i<len; i++) {
const VALUE index = ibf_load_small_value(load, &reading_pos);
rb_ary_push(ary, ibf_load_object(load, index));
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
if (header->internal) rb_obj_hide(ary);
if (header->frozen) rb_obj_freeze(ary);
return ary;
}
static int
ibf_dump_object_hash_i(st_data_t key, st_data_t val, st_data_t ptr)
{
struct ibf_dump *dump = (struct ibf_dump *)ptr;
VALUE key_index = ibf_dump_object(dump, (VALUE)key);
VALUE val_index = ibf_dump_object(dump, (VALUE)val);
ibf_dump_write_small_value(dump, key_index);
ibf_dump_write_small_value(dump, val_index);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
return ST_CONTINUE;
}
static void
ibf_dump_object_hash(struct ibf_dump *dump, VALUE obj)
{
long len = RHASH_SIZE(obj);
ibf_dump_write_small_value(dump, (VALUE)len);
if (len > 0) rb_hash_foreach(obj, ibf_dump_object_hash_i, (VALUE)dump);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
static VALUE
ibf_load_object_hash(const struct ibf_load *load, const struct ibf_object_header *header, ibf_offset_t offset)
{
long len = (long)ibf_load_small_value(load, &offset);
VALUE obj = rb_hash_new_with_size(len);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
int i;
for (i = 0; i < len; i++) {
VALUE key_index = ibf_load_small_value(load, &offset);
VALUE val_index = ibf_load_small_value(load, &offset);
VALUE key = ibf_load_object(load, key_index);
VALUE val = ibf_load_object(load, val_index);
rb_hash_aset(obj, key, val);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
rb_hash_rehash(obj);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
if (header->internal) rb_obj_hide(obj);
if (header->frozen) rb_obj_freeze(obj);
return obj;
}
static void
ibf_dump_object_struct(struct ibf_dump *dump, VALUE obj)
{
if (rb_obj_is_kind_of(obj, rb_cRange)) {
struct ibf_object_struct_range range;
VALUE beg, end;
IBF_ZERO(range);
range.len = 3;
range.class_index = 0;
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
rb_range_values(obj, &beg, &end, &range.excl);
range.beg = (long)ibf_dump_object(dump, beg);
range.end = (long)ibf_dump_object(dump, end);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
IBF_W_ALIGN(struct ibf_object_struct_range);
IBF_WV(range);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
else {
rb_raise(rb_eNotImpError, "ibf_dump_object_struct: unsupported class %"PRIsVALUE,
rb_class_name(CLASS_OF(obj)));
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
}
static VALUE
ibf_load_object_struct(const struct ibf_load *load, const struct ibf_object_header *header, ibf_offset_t offset)
{
const struct ibf_object_struct_range *range = IBF_OBJBODY(struct ibf_object_struct_range, offset);
VALUE beg = ibf_load_object(load, range->beg);
VALUE end = ibf_load_object(load, range->end);
VALUE obj = rb_range_new(beg, end, range->excl);
if (header->internal) rb_obj_hide(obj);
if (header->frozen) rb_obj_freeze(obj);
return obj;
}
static void
ibf_dump_object_bignum(struct ibf_dump *dump, VALUE obj)
{
ssize_t len = BIGNUM_LEN(obj);
ssize_t slen = BIGNUM_SIGN(obj) > 0 ? len : len * -1;
BDIGIT *d = BIGNUM_DIGITS(obj);
(void)IBF_W(&slen, ssize_t, 1);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
IBF_WP(d, BDIGIT, len);
}
static VALUE
ibf_load_object_bignum(const struct ibf_load *load, const struct ibf_object_header *header, ibf_offset_t offset)
{
const struct ibf_object_bignum *bignum = IBF_OBJBODY(struct ibf_object_bignum, offset);
int sign = bignum->slen > 0;
ssize_t len = sign > 0 ? bignum->slen : -1 * bignum->slen;
VALUE obj = rb_integer_unpack(bignum->digits, len * 2, 2, 0,
INTEGER_PACK_LITTLE_ENDIAN | (sign == 0 ? INTEGER_PACK_NEGATIVE : 0));
if (header->internal) rb_obj_hide(obj);
if (header->frozen) rb_obj_freeze(obj);
return obj;
}
static void
ibf_dump_object_data(struct ibf_dump *dump, VALUE obj)
{
if (rb_data_is_encoding(obj)) {
rb_encoding *enc = rb_to_encoding(obj);
const char *name = rb_enc_name(enc);
long len = strlen(name) + 1;
long data[2];
data[0] = IBF_OBJECT_DATA_ENCODING;
data[1] = len;
(void)IBF_W(data, long, 2);
IBF_WP(name, char, len);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
else {
ibf_dump_object_unsupported(dump, obj);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
}
static VALUE
ibf_load_object_data(const struct ibf_load *load, const struct ibf_object_header *header, ibf_offset_t offset)
{
const long *body = IBF_OBJBODY(long, offset);
const enum ibf_object_data_type type = (enum ibf_object_data_type)body[0];
/* const long len = body[1]; */
const char *data = (const char *)&body[2];
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
switch (type) {
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
case IBF_OBJECT_DATA_ENCODING:
{
VALUE encobj = rb_enc_from_encoding(rb_enc_find(data));
return encobj;
}
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
return ibf_load_object_unsupported(load, header, offset);
}
static void
ibf_dump_object_complex_rational(struct ibf_dump *dump, VALUE obj)
{
long data[2];
data[0] = (long)ibf_dump_object(dump, RCOMPLEX(obj)->real);
data[1] = (long)ibf_dump_object(dump, RCOMPLEX(obj)->imag);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
(void)IBF_W(data, long, 2);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
static VALUE
ibf_load_object_complex_rational(const struct ibf_load *load, const struct ibf_object_header *header, ibf_offset_t offset)
{
const struct ibf_object_complex_rational *nums = IBF_OBJBODY(struct ibf_object_complex_rational, offset);
VALUE a = ibf_load_object(load, nums->a);
VALUE b = ibf_load_object(load, nums->b);
VALUE obj = header->type == T_COMPLEX ?
rb_complex_new(a, b) : rb_rational_new(a, b);
if (header->internal) rb_obj_hide(obj);
if (header->frozen) rb_obj_freeze(obj);
return obj;
}
static void
ibf_dump_object_symbol(struct ibf_dump *dump, VALUE obj)
{
VALUE str = rb_sym2str(obj);
VALUE str_index = ibf_dump_object(dump, str);
ibf_dump_write_small_value(dump, str_index);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
static VALUE
ibf_load_object_symbol(const struct ibf_load *load, const struct ibf_object_header *header, ibf_offset_t offset)
{
VALUE str_index = ibf_load_small_value(load, &offset);
VALUE str = ibf_load_object(load, str_index);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
ID id = rb_intern_str(str);
return ID2SYM(id);
}
typedef void (*ibf_dump_object_function)(struct ibf_dump *dump, VALUE obj);
static ibf_dump_object_function dump_object_functions[RUBY_T_MASK+1] = {
ibf_dump_object_unsupported, /* T_NONE */
ibf_dump_object_unsupported, /* T_OBJECT */
ibf_dump_object_class, /* T_CLASS */
ibf_dump_object_unsupported, /* T_MODULE */
ibf_dump_object_float, /* T_FLOAT */
ibf_dump_object_string, /* T_STRING */
ibf_dump_object_regexp, /* T_REGEXP */
ibf_dump_object_array, /* T_ARRAY */
ibf_dump_object_hash, /* T_HASH */
ibf_dump_object_struct, /* T_STRUCT */
ibf_dump_object_bignum, /* T_BIGNUM */
ibf_dump_object_unsupported, /* T_FILE */
ibf_dump_object_data, /* T_DATA */
ibf_dump_object_unsupported, /* T_MATCH */
ibf_dump_object_complex_rational, /* T_COMPLEX */
ibf_dump_object_complex_rational, /* T_RATIONAL */
ibf_dump_object_unsupported, /* 0x10 */
ibf_dump_object_unsupported, /* 0x11 T_NIL */
ibf_dump_object_unsupported, /* 0x12 T_TRUE */
ibf_dump_object_unsupported, /* 0x13 T_FALSE */
ibf_dump_object_symbol, /* 0x14 T_SYMBOL */
ibf_dump_object_unsupported, /* T_FIXNUM */
ibf_dump_object_unsupported, /* T_UNDEF */
ibf_dump_object_unsupported, /* 0x17 */
ibf_dump_object_unsupported, /* 0x18 */
ibf_dump_object_unsupported, /* 0x19 */
ibf_dump_object_unsupported, /* T_IMEMO 0x1a */
ibf_dump_object_unsupported, /* T_NODE 0x1b */
ibf_dump_object_unsupported, /* T_ICLASS 0x1c */
ibf_dump_object_unsupported, /* T_ZOMBIE 0x1d */
ibf_dump_object_unsupported, /* 0x1e */
ibf_dump_object_unsupported, /* 0x1f */
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
};
static void
ibf_dump_object_object_header(struct ibf_dump *dump, const struct ibf_object_header header)
{
unsigned char byte =
(header.type << 0) |
(header.special_const << 5) |
(header.frozen << 6) |
(header.internal << 7);
IBF_WV(byte);
}
static struct ibf_object_header
ibf_load_object_object_header(const struct ibf_load *load, ibf_offset_t *offset)
{
unsigned char byte = ibf_load_byte(load, offset);
struct ibf_object_header header;
header.type = (byte >> 0) & 0x1f;
header.special_const = (byte >> 5) & 0x01;
header.frozen = (byte >> 6) & 0x01;
header.internal = (byte >> 7) & 0x01;
return header;
}
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
static ibf_offset_t
ibf_dump_object_object(struct ibf_dump *dump, VALUE obj_list, VALUE obj)
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
{
struct ibf_object_header obj_header;
ibf_offset_t current_offset;
IBF_ZERO(obj_header);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
obj_header.type = TYPE(obj);
IBF_W_ALIGN(ibf_offset_t);
current_offset = ibf_dump_pos(dump);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
if (SPECIAL_CONST_P(obj)) {
if (RB_TYPE_P(obj, T_SYMBOL) ||
RB_TYPE_P(obj, T_FLOAT)) {
obj_header.internal = FALSE;
goto dump_object;
}
obj_header.special_const = TRUE;
obj_header.frozen = TRUE;
obj_header.internal = TRUE;
ibf_dump_object_object_header(dump, obj_header);
ibf_dump_write_small_value(dump, obj);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
else {
obj_header.internal = (RBASIC_CLASS(obj) == 0) ? TRUE : FALSE;
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
dump_object:
obj_header.special_const = FALSE;
obj_header.frozen = FL_TEST(obj, FL_FREEZE) ? TRUE : FALSE;
ibf_dump_object_object_header(dump, obj_header);
(*dump_object_functions[obj_header.type])(dump, obj);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
return current_offset;
}
typedef VALUE (*ibf_load_object_function)(const struct ibf_load *load, const struct ibf_object_header *header, ibf_offset_t offset);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
static ibf_load_object_function load_object_functions[RUBY_T_MASK+1] = {
ibf_load_object_unsupported, /* T_NONE */
ibf_load_object_unsupported, /* T_OBJECT */
ibf_load_object_class, /* T_CLASS */
ibf_load_object_unsupported, /* T_MODULE */
ibf_load_object_float, /* T_FLOAT */
ibf_load_object_string, /* T_STRING */
ibf_load_object_regexp, /* T_REGEXP */
ibf_load_object_array, /* T_ARRAY */
ibf_load_object_hash, /* T_HASH */
ibf_load_object_struct, /* T_STRUCT */
ibf_load_object_bignum, /* T_BIGNUM */
ibf_load_object_unsupported, /* T_FILE */
ibf_load_object_data, /* T_DATA */
ibf_load_object_unsupported, /* T_MATCH */
ibf_load_object_complex_rational, /* T_COMPLEX */
ibf_load_object_complex_rational, /* T_RATIONAL */
ibf_load_object_unsupported, /* 0x10 */
ibf_load_object_unsupported, /* T_NIL */
ibf_load_object_unsupported, /* T_TRUE */
ibf_load_object_unsupported, /* T_FALSE */
ibf_load_object_symbol,
ibf_load_object_unsupported, /* T_FIXNUM */
ibf_load_object_unsupported, /* T_UNDEF */
ibf_load_object_unsupported, /* 0x17 */
ibf_load_object_unsupported, /* 0x18 */
ibf_load_object_unsupported, /* 0x19 */
ibf_load_object_unsupported, /* T_IMEMO 0x1a */
ibf_load_object_unsupported, /* T_NODE 0x1b */
ibf_load_object_unsupported, /* T_ICLASS 0x1c */
ibf_load_object_unsupported, /* T_ZOMBIE 0x1d */
ibf_load_object_unsupported, /* 0x1e */
ibf_load_object_unsupported, /* 0x1f */
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
};
static VALUE
ibf_load_object(const struct ibf_load *load, VALUE object_index)
{
if (object_index == 0) {
return Qnil;
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
else if (object_index >= (VALUE)RARRAY_LEN(load->current_buffer->obj_list)) {
rb_raise(rb_eIndexError, "object index out of range: %"PRIdVALUE, object_index);
}
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
else {
VALUE obj = rb_ary_entry(load->current_buffer->obj_list, (long)object_index);
if (obj == Qnil) { /* TODO: avoid multiple Qnil load */
ibf_offset_t *offsets = (ibf_offset_t *)(load->current_buffer->obj_list_offset + load->current_buffer->buff);
ibf_offset_t offset = offsets[object_index];
const struct ibf_object_header header = ibf_load_object_object_header(load, &offset);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
#if IBF_ISEQ_DEBUG
fprintf(stderr, "ibf_load_object: list=%#x offsets=%p offset=%#x\n",
load->current_buffer->obj_list_offset, offsets, offset);
fprintf(stderr, "ibf_load_object: type=%#x special=%d frozen=%d internal=%d\n",
header.type, header.special_const, header.frozen, header.internal);
#endif
if (offset >= load->current_buffer->size) {
rb_raise(rb_eIndexError, "object offset out of range: %u", offset);
}
if (header.special_const) {
ibf_offset_t reading_pos = offset;
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
obj = ibf_load_small_value(load, &reading_pos);
}
else {
obj = (*load_object_functions[header.type])(load, &header, offset);
}
rb_ary_store(load->current_buffer->obj_list, (long)object_index, obj);
}
#if IBF_ISEQ_DEBUG
fprintf(stderr, "ibf_load_object: index=%#"PRIxVALUE" obj=%#"PRIxVALUE"\n",
object_index, obj);
#endif
return obj;
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
}
static void
ibf_dump_object_list(struct ibf_dump *dump, ibf_offset_t *obj_list_offset, unsigned int *obj_list_size)
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
{
VALUE obj_list = dump->current_buffer->obj_list;
VALUE list = rb_ary_tmp_new(RARRAY_LEN(obj_list));
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
int i, size;
for (i=0; i<RARRAY_LEN(obj_list); i++) {
VALUE obj = RARRAY_AREF(obj_list, i);
ibf_offset_t offset = ibf_dump_object_object(dump, obj_list, obj);
rb_ary_push(list, UINT2NUM(offset));
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
size = i;
IBF_W_ALIGN(ibf_offset_t);
*obj_list_offset = ibf_dump_pos(dump);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
for (i=0; i<size; i++) {
ibf_offset_t offset = NUM2UINT(RARRAY_AREF(list, i));
IBF_WV(offset);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
*obj_list_size = size;
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
static void
ibf_dump_mark(void *ptr)
{
struct ibf_dump *dump = (struct ibf_dump *)ptr;
rb_gc_mark(dump->global_buffer.str);
rb_gc_mark(dump->global_buffer.obj_list);
rb_gc_mark(dump->iseq_list);
}
static void
ibf_dump_free(void *ptr)
{
struct ibf_dump *dump = (struct ibf_dump *)ptr;
if (dump->iseq_table) {
st_free_table(dump->iseq_table);
dump->iseq_table = 0;
}
ruby_xfree(dump);
}
static size_t
ibf_dump_memsize(const void *ptr)
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
{
struct ibf_dump *dump = (struct ibf_dump *)ptr;
size_t size = sizeof(*dump);
if (dump->iseq_table) size += st_memsize(dump->iseq_table);
return size;
}
static const rb_data_type_t ibf_dump_type = {
"ibf_dump",
{ibf_dump_mark, ibf_dump_free, ibf_dump_memsize,},
0, 0, RUBY_TYPED_WB_PROTECTED | RUBY_TYPED_FREE_IMMEDIATELY
};
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
static void
ibf_dump_setup(struct ibf_dump *dump, VALUE dumper_obj)
{
RB_OBJ_WRITE(dumper_obj, &dump->iseq_list, rb_ary_tmp_new(0));
RB_OBJ_WRITE(dumper_obj, &dump->global_buffer.obj_list, ibf_dump_object_list_new());
RB_OBJ_WRITE(dumper_obj, &dump->global_buffer.str, rb_str_new(0, 0));
dump->iseq_table = st_init_numtable(); /* need free */
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
dump->current_buffer = &dump->global_buffer;
}
VALUE
rb_iseq_ibf_dump(const rb_iseq_t *iseq, VALUE opt)
{
struct ibf_dump *dump;
struct ibf_header header = {{0}};
VALUE dump_obj;
VALUE str;
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
if (iseq->body->parent_iseq != NULL ||
iseq->body->local_iseq != iseq) {
rb_raise(rb_eRuntimeError, "should be top of iseq");
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
if (RTEST(ISEQ_COVERAGE(iseq))) {
rb_raise(rb_eRuntimeError, "should not compile with coverage");
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
dump_obj = TypedData_Make_Struct(0, struct ibf_dump, &ibf_dump_type, dump);
ibf_dump_setup(dump, dump_obj);
ibf_dump_write(dump, &header, sizeof(header));
ibf_dump_write(dump, RUBY_PLATFORM, strlen(RUBY_PLATFORM) + 1);
ibf_dump_iseq(dump, iseq);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
header.magic[0] = 'Y'; /* YARB */
header.magic[1] = 'A';
header.magic[2] = 'R';
header.magic[3] = 'B';
header.major_version = IBF_MAJOR_VERSION;
header.minor_version = IBF_MINOR_VERSION;
ibf_dump_iseq_list(dump, &header);
ibf_dump_object_list(dump, &header.global_object_list_offset, &header.global_object_list_size);
header.size = ibf_dump_pos(dump);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
if (RTEST(opt)) {
VALUE opt_str = opt;
const char *ptr = StringValuePtr(opt_str);
header.extra_size = RSTRING_LENINT(opt_str);
ibf_dump_write(dump, ptr, header.extra_size);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
else {
header.extra_size = 0;
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
ibf_dump_overwrite(dump, &header, sizeof(header), 0);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
str = dump->global_buffer.str;
ibf_dump_free(dump);
DATA_PTR(dump_obj) = NULL;
RB_GC_GUARD(dump_obj);
return str;
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
static const ibf_offset_t *
ibf_iseq_list(const struct ibf_load *load)
{
return (const ibf_offset_t *)(load->global_buffer.buff + load->header->iseq_list_offset);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
void
rb_ibf_load_iseq_complete(rb_iseq_t *iseq)
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
{
struct ibf_load *load = RTYPEDDATA_DATA(iseq->aux.loader.obj);
rb_iseq_t *prev_src_iseq = load->iseq;
ibf_offset_t offset = ibf_iseq_list(load)[iseq->aux.loader.index];
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
load->iseq = iseq;
#if IBF_ISEQ_DEBUG
fprintf(stderr, "rb_ibf_load_iseq_complete: index=%#x offset=%#x size=%#x\n",
iseq->aux.loader.index, offset,
load->header->size);
#endif
ibf_load_iseq_each(load, iseq, offset);
ISEQ_COMPILE_DATA_CLEAR(iseq);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
FL_UNSET(iseq, ISEQ_NOT_LOADED_YET);
rb_iseq_init_trace(iseq);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
load->iseq = prev_src_iseq;
}
#if USE_LAZY_LOAD
const rb_iseq_t *
rb_iseq_complete(const rb_iseq_t *iseq)
{
rb_ibf_load_iseq_complete((rb_iseq_t *)iseq);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
return iseq;
}
#endif
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
static rb_iseq_t *
ibf_load_iseq(const struct ibf_load *load, const rb_iseq_t *index_iseq)
{
int iseq_index = (int)(VALUE)index_iseq;
#if IBF_ISEQ_DEBUG
fprintf(stderr, "ibf_load_iseq: index_iseq=%p iseq_list=%p\n",
index_iseq, (void *)load->iseq_list);
#endif
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
if (iseq_index == -1) {
return NULL;
}
else {
VALUE iseqv = rb_ary_entry(load->iseq_list, iseq_index);
#if IBF_ISEQ_DEBUG
fprintf(stderr, "ibf_load_iseq: iseqv=%p\n", (void *)iseqv);
#endif
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
if (iseqv != Qnil) {
return (rb_iseq_t *)iseqv;
}
else {
rb_iseq_t *iseq = iseq_imemo_alloc();
#if IBF_ISEQ_DEBUG
fprintf(stderr, "ibf_load_iseq: new iseq=%p\n", iseq);
#endif
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
FL_SET(iseq, ISEQ_NOT_LOADED_YET);
iseq->aux.loader.obj = load->loader_obj;
iseq->aux.loader.index = iseq_index;
#if IBF_ISEQ_DEBUG
fprintf(stderr, "ibf_load_iseq: iseq=%p loader_obj=%p index=%d\n",
iseq, (void *)load->loader_obj, iseq_index);
#endif
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
rb_ary_store(load->iseq_list, iseq_index, (VALUE)iseq);
#if !USE_LAZY_LOAD
#if IBF_ISEQ_DEBUG
fprintf(stderr, "ibf_load_iseq: loading iseq=%p\n", iseq);
#endif
rb_ibf_load_iseq_complete(iseq);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
#endif /* !USE_LAZY_LOAD */
#if IBF_ISEQ_DEBUG
fprintf(stderr, "ibf_load_iseq: iseq=%p loaded %p\n",
iseq, load->iseq);
#endif
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
return iseq;
}
}
}
static void
ibf_load_setup(struct ibf_load *load, VALUE loader_obj, VALUE str)
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
{
rb_check_safe_obj(str);
if (RSTRING_LENINT(str) < (int)sizeof(struct ibf_header)) {
rb_raise(rb_eRuntimeError, "broken binary format");
}
#if USE_LAZY_LOAD
str = rb_str_new(RSTRING_PTR(str), RSTRING_LEN(str));
#endif
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
RB_OBJ_WRITE(loader_obj, &load->str, str);
load->loader_obj = loader_obj;
load->global_buffer.buff = StringValuePtr(str);
load->header = (struct ibf_header *)load->global_buffer.buff;
load->global_buffer.size = load->header->size;
load->global_buffer.obj_list_offset = load->header->global_object_list_offset;
load->global_buffer.obj_list_size = load->header->global_object_list_size;
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
RB_OBJ_WRITE(loader_obj, &load->iseq_list, rb_ary_tmp_new(0));
RB_OBJ_WRITE(loader_obj, &load->global_buffer.obj_list, rb_ary_tmp_new(load->global_buffer.obj_list_size));
rb_ary_resize(load->global_buffer.obj_list, load->global_buffer.obj_list_size);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
load->iseq = NULL;
load->current_buffer = &load->global_buffer;
if (RSTRING_LENINT(str) < (int)load->header->size) {
rb_raise(rb_eRuntimeError, "broken binary format");
}
if (strncmp(load->header->magic, "YARB", 4) != 0) {
rb_raise(rb_eRuntimeError, "unknown binary format");
}
if (load->header->major_version != IBF_MAJOR_VERSION ||
load->header->minor_version != IBF_MINOR_VERSION) {
rb_raise(rb_eRuntimeError, "unmatched version file (%u.%u for %u.%u)",
load->header->major_version, load->header->minor_version, IBF_MAJOR_VERSION, IBF_MINOR_VERSION);
}
if (strcmp(load->global_buffer.buff + sizeof(struct ibf_header), RUBY_PLATFORM) != 0) {
rb_raise(rb_eRuntimeError, "unmatched platform");
}
if (load->header->iseq_list_offset % RUBY_ALIGNOF(ibf_offset_t)) {
rb_raise(rb_eArgError, "unaligned iseq list offset: %u",
load->header->iseq_list_offset);
}
if (load->global_buffer.obj_list_offset % RUBY_ALIGNOF(ibf_offset_t)) {
rb_raise(rb_eArgError, "unaligned object list offset: %u",
load->global_buffer.obj_list_offset);
}
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
static void
ibf_loader_mark(void *ptr)
{
struct ibf_load *load = (struct ibf_load *)ptr;
rb_gc_mark(load->str);
rb_gc_mark(load->iseq_list);
rb_gc_mark(load->global_buffer.obj_list);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
static void
ibf_loader_free(void *ptr)
{
struct ibf_load *load = (struct ibf_load *)ptr;
ruby_xfree(load);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
static size_t
ibf_loader_memsize(const void *ptr)
{
return sizeof(struct ibf_load);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
}
static const rb_data_type_t ibf_load_type = {
"ibf_loader",
{ibf_loader_mark, ibf_loader_free, ibf_loader_memsize,},
0, 0, RUBY_TYPED_WB_PROTECTED | RUBY_TYPED_FREE_IMMEDIATELY
};
const rb_iseq_t *
rb_iseq_ibf_load(VALUE str)
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
{
struct ibf_load *load;
rb_iseq_t *iseq;
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
VALUE loader_obj = TypedData_Make_Struct(0, struct ibf_load, &ibf_load_type, load);
ibf_load_setup(load, loader_obj, str);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
iseq = ibf_load_iseq(load, 0);
RB_GC_GUARD(loader_obj);
return iseq;
}
VALUE
rb_iseq_ibf_load_extra_data(VALUE str)
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
{
struct ibf_load *load;
VALUE loader_obj = TypedData_Make_Struct(0, struct ibf_load, &ibf_load_type, load);
VALUE extra_str;
ibf_load_setup(load, loader_obj, str);
extra_str = rb_str_new(load->global_buffer.buff + load->header->size, load->header->extra_size);
* introduce new ISeq binary format serializer/de-serializer and a pre-compilation/runtime loader sample. [Feature #11788] * iseq.c: add new methods: * RubyVM::InstructionSequence#to_binary_format(extra_data = nil) * RubyVM::InstructionSequence.from_binary_format(binary) * RubyVM::InstructionSequence.from_binary_format_extra_data(binary) * compile.c: implement body of this new feature. * load.c (rb_load_internal0), iseq.c (rb_iseq_load_iseq): call RubyVM::InstructionSequence.load_iseq(fname) with loading script name if this method is defined. We can return any ISeq object as a result value. Otherwise loading will be continue as usual. This interface is not matured and is not extensible. So that we don't guarantee the future compatibility of this method. Basically, you should'nt use this method. * iseq.h: move ISEQ_MAJOR/MINOR_VERSION (and some definitions) from iseq.c. * encoding.c (rb_data_is_encoding), internal.h: added. * vm_core.h: add several supports for lazy load. * add USE_LAZY_LOAD macro to specify enable or disable of this feature. * add several fields to rb_iseq_t. * introduce new macro rb_iseq_check(). * insns.def: some check for lazy loading feature. * vm_insnhelper.c: ditto. * proc.c: ditto. * vm.c: ditto. * test/lib/iseq_loader_checker.rb: enabled iff suitable environment variables are provided. * test/runner.rb: enable lib/iseq_loader_checker.rb. * sample/iseq_loader.rb: add sample compiler and loader. $ ruby sample/iseq_loader.rb [dir] will compile all ruby scripts in [dir]. With default setting, this compile creates *.rb.yarb files in same directory of target .rb scripts. $ ruby -r sample/iseq_loader.rb [app] will run with enable to load compiled binary data. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@52949 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
2015-12-08 16:58:50 +03:00
RB_GC_GUARD(loader_obj);
return extra_str;
}