Speedup block.call [Feature #14330]

* insns.def (getblockparamproxy): introduce new instruction to return
  the `rb_block_param_proxy` object if possible. This object responds
  to `call` method and invoke given block (completely similar to `yield`).

* method.h (OPTIMIZED_METHOD_TYPE_BLOCK_CALL): add new optimized call type
  which is for `rb_block_param_proxy.cal`.

* vm_insnhelper.c (vm_call_method_each_type): ditto.

* vm_insnhelper.c (vm_call_opt_block_call): ditto.

* vm_core.h (BOP_CALL, PROC_REDEFINED_OP_FLAG): add check for `Proc#call`
  redefinition.

* compile.c (iseq_compile_each0): compile to use new insn
  `getblockparamproxy` for method call.


git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@61659 b2dd03c8-39d4-4d8f-98ff-823fe69b080e

compile.c

@@ -1399,6 +1399,21 @@ iseq_local_block_param_p(const rb_iseq_t *iseq, unsigned int idx, unsigned int l
     }
 }
 
+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)
 {
@@ -6159,7 +6174,17 @@ iseq_compile_each0(rb_iseq_t *iseq, LINK_ANCHOR *const ret, const NODE *node, in
 #endif
 	/* receiver */
 	if (type == NODE_CALL || type == NODE_OPCALL || type == NODE_QCALL) {
-	    CHECK(COMPILE(recv, "recv", node->nd_recv));
+	    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 {
+		CHECK(COMPILE(recv, "recv", node->nd_recv));
+	    }
+
 	    if (type == NODE_QCALL) {
 		else_label = NEW_LABEL(line);
 		end_label = NEW_LABEL(line);

insns.def

@@ -127,6 +127,54 @@ setblockparam
     VM_ENV_FLAGS_SET(ep, VM_FRAME_FLAG_MODIFIED_BLOCK_PARAM);
 }
 
+/**
+  @c variable
+  @e Get special proxy object which only responds to `call` method if the block parameter
+     represents a iseq/ifunc block. Otherwise, same as `getblockparam`.
+  @j ブロックパラメータが iseq/ifunc ブロックであれば、特殊なプロキシオブジェクトを取得する。
+ */
+DEFINE_INSN
+getblockparamproxy
+(lindex_t idx, rb_num_t level)
+()
+(VALUE val)
+{
+    const VALUE *ep = vm_get_ep(GET_EP(), level);
+    VM_ASSERT(VM_ENV_LOCAL_P(ep));
+
+    if (!VM_ENV_FLAGS(ep, VM_FRAME_FLAG_MODIFIED_BLOCK_PARAM)) {
+	VALUE block_handler = VM_ENV_BLOCK_HANDLER(ep);
+
+	if (block_handler) {
+	    switch (vm_block_handler_type(block_handler)) {
+	      case block_handler_type_iseq:
+	      case block_handler_type_ifunc:
+		val = rb_block_param_proxy;
+		break;
+	      case block_handler_type_symbol:
+		val = rb_sym_to_proc(VM_BH_TO_SYMBOL(block_handler));
+		goto INSN_LABEL(set);
+	      case block_handler_type_proc:
+		val = VM_BH_TO_PROC(block_handler);
+		goto INSN_LABEL(set);
+	      default:
+		VM_UNREACHABLE(getblockparamproxy);
+	    }
+	}
+	else {
+	    val = Qnil;
+	  INSN_LABEL(set):
+	    vm_env_write(ep, -(int)idx, val);
+	    VM_ENV_FLAGS_SET(ep, VM_FRAME_FLAG_MODIFIED_BLOCK_PARAM);
+	}
+    }
+    else {
+	val = *(ep - idx);
+	RB_DEBUG_COUNTER_INC(lvar_get);
+	(void)RB_DEBUG_COUNTER_INC_IF(lvar_get_dynamic, level > 0);
+    }
+}
+
 /**
   @c variable
   @e Get value of special local variable ($~, $_, ..).

method.h

@@ -154,6 +154,7 @@ typedef struct rb_method_refined_struct {
 enum method_optimized_type {
     OPTIMIZED_METHOD_TYPE_SEND,
     OPTIMIZED_METHOD_TYPE_CALL,
+    OPTIMIZED_METHOD_TYPE_BLOCK_CALL,
     OPTIMIZED_METHOD_TYPE__MAX
 };
 

proc.c

@@ -2329,6 +2329,9 @@ rb_method_entry_min_max_arity(const rb_method_entry_t *me, int *max)
 	  case OPTIMIZED_METHOD_TYPE_CALL:
 	    *max = UNLIMITED_ARGUMENTS;
 	    return 0;
+	  case OPTIMIZED_METHOD_TYPE_BLOCK_CALL:
+	    *max = UNLIMITED_ARGUMENTS;
+	    return 0;
 	  default:
 	    break;
 	}

vm.c

@@ -295,6 +295,8 @@ static VALUE vm_make_env_object(const rb_execution_context_t *ec, rb_control_fra
 static VALUE vm_invoke_bmethod(rb_execution_context_t *ec, rb_proc_t *proc, VALUE self, int argc, const VALUE *argv, VALUE block_handler);
 static VALUE vm_invoke_proc(rb_execution_context_t *ec, rb_proc_t *proc, VALUE self, int argc, const VALUE *argv, VALUE block_handler);
 
+static VALUE rb_block_param_proxy;
+
 #include "vm_insnhelper.h"
 #include "vm_exec.h"
 #include "vm_insnhelper.c"
@@ -1494,6 +1496,7 @@ vm_redefinition_check_flag(VALUE klass)
     if (klass == rb_cNilClass) return NIL_REDEFINED_OP_FLAG;
     if (klass == rb_cTrueClass) return TRUE_REDEFINED_OP_FLAG;
     if (klass == rb_cFalseClass) return FALSE_REDEFINED_OP_FLAG;
+    if (klass == rb_cProc) return PROC_REDEFINED_OP_FLAG;
     return 0;
 }
 
@@ -1537,7 +1540,9 @@ add_opt_method(VALUE klass, ID mid, VALUE bop)
 {
     const rb_method_entry_t *me = rb_method_entry_at(klass, mid);
 
-    if (me && me->def->type == VM_METHOD_TYPE_CFUNC) {
+    if (me &&
+	(me->def->type == VM_METHOD_TYPE_CFUNC ||
+	 me->def->type == VM_METHOD_TYPE_OPTIMIZED)) {
 	st_insert(vm_opt_method_table, (st_data_t)me, (st_data_t)bop);
     }
     else {
@@ -3054,6 +3059,12 @@ Init_VM(void)
     }
     vm_init_redefined_flag();
 
+    rb_block_param_proxy = rb_obj_alloc(rb_cObject);
+    rb_add_method(rb_singleton_class(rb_block_param_proxy), rb_intern("call"), VM_METHOD_TYPE_OPTIMIZED,
+		  (void *)OPTIMIZED_METHOD_TYPE_BLOCK_CALL, METHOD_VISI_PUBLIC);
+    rb_obj_freeze(rb_block_param_proxy);
+    rb_gc_register_mark_object(rb_block_param_proxy);
+
     /* vm_backtrace.c */
     Init_vm_backtrace();
     VM_PROFILE_ATEXIT();

vm_core.h

@@ -483,6 +483,7 @@ enum ruby_basic_operators {
     BOP_UMINUS,
     BOP_MAX,
     BOP_MIN,
+    BOP_CALL,
 
     BOP_LAST_
 };
@@ -619,6 +620,7 @@ typedef struct rb_vm_struct {
 #define NIL_REDEFINED_OP_FLAG    (1 << 9)
 #define TRUE_REDEFINED_OP_FLAG   (1 << 10)
 #define FALSE_REDEFINED_OP_FLAG  (1 << 11)
+#define PROC_REDEFINED_OP_FLAG   (1 << 12)
 
 #define BASIC_OP_UNREDEFINED_P(op, klass) (LIKELY((GET_VM()->redefined_flag[(op)]&(klass)) == 0))
 

vm_insnhelper.c

@@ -2047,7 +2047,18 @@ vm_call_opt_send(rb_execution_context_t *ec, rb_control_frame_t *reg_cfp, struct
     return vm_call_method(ec, reg_cfp, calling, ci, cc);
 }
 
-static VALUE vm_invoke_block(rb_execution_context_t *ec, rb_control_frame_t *reg_cfp, struct rb_calling_info *calling, const struct rb_call_info *ci, VALUE block_handler);
+static inline VALUE vm_invoke_block(rb_execution_context_t *ec, rb_control_frame_t *reg_cfp, struct rb_calling_info *calling, const struct rb_call_info *ci, VALUE block_handler);
+
+NOINLINE(static VALUE
+	 vm_invoke_block_noinline(rb_execution_context_t *ec, rb_control_frame_t *reg_cfp,
+				  struct rb_calling_info *calling, const struct rb_call_info *ci, VALUE block_handler));
+
+static VALUE
+vm_invoke_block_noinline(rb_execution_context_t *ec, rb_control_frame_t *reg_cfp,
+				      struct rb_calling_info *calling, const struct rb_call_info *ci, VALUE block_handler)
+{
+    return vm_invoke_block(ec, reg_cfp, calling, ci, block_handler);
+}
 
 static VALUE
 vm_call_opt_call(rb_execution_context_t *ec, rb_control_frame_t *reg_cfp, struct rb_calling_info *calling, const struct rb_call_info *ci, struct rb_call_cache *cc)
@@ -2059,7 +2070,24 @@ vm_call_opt_call(rb_execution_context_t *ec, rb_control_frame_t *reg_cfp, struct
     if (argc > 0) MEMMOVE(&TOPN(argc), &TOPN(argc-1), VALUE, argc);
     DEC_SP(1);
 
-    return vm_invoke_block(ec, reg_cfp, calling, ci, VM_BH_FROM_PROC(procval));
+    return vm_invoke_block_noinline(ec, reg_cfp, calling, ci, VM_BH_FROM_PROC(procval));
+}
+
+static VALUE
+vm_call_opt_block_call(rb_execution_context_t *ec, rb_control_frame_t *reg_cfp, struct rb_calling_info *calling, const struct rb_call_info *ci, struct rb_call_cache *cc)
+{
+    int argc = calling->argc;
+    VALUE block_handler = VM_ENV_BLOCK_HANDLER(reg_cfp->ep);
+
+    if (BASIC_OP_UNREDEFINED_P(BOP_CALL, PROC_REDEFINED_OP_FLAG)) {
+	if (argc > 0) MEMMOVE(&TOPN(argc), &TOPN(argc-1), VALUE, argc);
+	DEC_SP(1);
+	return vm_invoke_block_noinline(ec, reg_cfp, calling, ci, block_handler);
+    }
+    else {
+	calling->recv = rb_vm_bh_to_procval(ec, block_handler);
+	return vm_call_general(ec, reg_cfp, calling, ci, cc);
+    }
 }
 
 static VALUE
@@ -2264,6 +2292,9 @@ vm_call_method_each_type(rb_execution_context_t *ec, rb_control_frame_t *cfp, st
 	  case OPTIMIZED_METHOD_TYPE_CALL:
 	    CI_SET_FASTPATH(cc, vm_call_opt_call, TRUE);
 	    return vm_call_opt_call(ec, cfp, calling, ci, cc);
+	  case OPTIMIZED_METHOD_TYPE_BLOCK_CALL:
+	    CI_SET_FASTPATH(cc, vm_call_opt_block_call, TRUE);
+	    return vm_call_opt_block_call(ec, cfp, calling, ci, cc);
 	  default:
 	    rb_bug("vm_call_method: unsupported optimized method type (%d)",
 		   cc->me->def->body.optimize_type);
@@ -2689,8 +2720,9 @@ vm_proc_to_block_handler(VALUE procval)
     return Qundef;
 }
 
-static VALUE
-vm_invoke_block(rb_execution_context_t *ec, rb_control_frame_t *reg_cfp, struct rb_calling_info *calling, const struct rb_call_info *ci, VALUE block_handler)
+static inline VALUE
+vm_invoke_block(rb_execution_context_t *ec, rb_control_frame_t *reg_cfp,
+		struct rb_calling_info *calling, const struct rb_call_info *ci, VALUE block_handler)
 {
     int is_lambda = FALSE;