2 * Copyright (C) 1995-2011 University of Karlsruhe. All right reserved.
4 * This file is part of libFirm.
6 * This file may be distributed and/or modified under the terms of the
7 * GNU General Public License version 2 as published by the Free Software
8 * Foundation and appearing in the file LICENSE.GPL included in the
9 * packaging of this file.
11 * Licensees holding valid libFirm Professional Edition licenses may use
12 * this file in accordance with the libFirm Commercial License.
13 * Agreement provided with the Software.
15 * This file is provided AS IS with NO WARRANTY OF ANY KIND, INCLUDING THE
16 * WARRANTY OF DESIGN, MERCHANTABILITY AND FITNESS FOR A PARTICULAR
22 * @brief Dead node elimination and Procedure Inlining.
23 * @author Michael Beck, Goetz Lindenmaier
32 #include "irgraph_t.h"
35 #include "iroptimize.h"
52 #include "irbackedge_t.h"
58 #include "analyze_irg_args.h"
59 #include "iredges_t.h"
63 #include "iropt_dbg.h"
65 #include "irnodemap.h"
67 DEBUG_ONLY(static firm_dbg_module_t *dbg;)
69 /*------------------------------------------------------------------*/
70 /* Routines for dead node elimination / copying garbage collection */
72 /*------------------------------------------------------------------*/
75 * Remember the new node in the old node by using a field all nodes have.
77 static void set_new_node(ir_node *node, ir_node *new_node)
79 set_irn_link(node, new_node);
83 * Get this new node, before the old node is forgotten.
85 static inline ir_node *get_new_node(ir_node *old_node)
87 assert(irn_visited(old_node));
88 return (ir_node*) get_irn_link(old_node);
91 /*--------------------------------------------------------------------*/
92 /* Functionality for inlining */
93 /*--------------------------------------------------------------------*/
96 * Copy node for inlineing. Updates attributes that change when
97 * inlineing but not for dead node elimination.
99 * Copies the node by calling copy_node() and then updates the entity if
100 * it's a local one. env must be a pointer of the frame type of the
101 * inlined procedure. The new entities must be in the link field of
104 static void copy_node_inline(ir_node *node, void *env)
106 ir_graph *new_irg = (ir_graph*) env;
107 ir_node *new_node = irn_copy_into_irg(node, new_irg);
109 set_new_node(node, new_node);
111 ir_graph *old_irg = get_irn_irg(node);
112 ir_type *old_frame_type = get_irg_frame_type(old_irg);
113 ir_entity *old_entity = get_Sel_entity(node);
114 assert(is_Sel(new_node));
115 /* use copied entities from the new frame */
116 if (get_entity_owner(old_entity) == old_frame_type) {
117 ir_entity *new_entity = (ir_entity*)get_entity_link(old_entity);
118 assert(new_entity != NULL);
119 set_Sel_entity(new_node, new_entity);
121 } else if (is_Block(new_node)) {
122 new_node->attr.block.irg.irg = new_irg;
126 static void set_preds_inline(ir_node *node, void *env)
130 irn_rewire_inputs(node);
132 /* move constants into start block */
133 new_node = get_new_node(node);
134 if (is_irn_constlike(new_node)) {
135 ir_graph *new_irg = (ir_graph *) env;
136 ir_node *start_block = get_irg_start_block(new_irg);
137 set_nodes_block(new_node, start_block);
142 * Walker: checks if P_value_arg_base is used.
144 static void find_addr(ir_node *node, void *env)
146 bool *allow_inline = (bool*)env;
148 if (is_Block(node) && get_Block_entity(node)) {
150 * Currently we can't handle blocks whose address was taken correctly
153 *allow_inline = false;
154 } else if (is_Sel(node)) {
155 ir_graph *irg = current_ir_graph;
156 if (get_Sel_ptr(node) == get_irg_frame(irg)) {
157 /* access to frame */
158 ir_entity *ent = get_Sel_entity(node);
159 if (get_entity_owner(ent) != get_irg_frame_type(irg)) {
160 /* access to value_type */
161 *allow_inline = false;
163 if (is_parameter_entity(ent)) {
164 *allow_inline = false;
167 } else if (is_Alloc(node) && get_Alloc_where(node) == stack_alloc) {
169 * Refuse to inline alloca call unless user explicitly forced so as this
170 * may change program's memory overhead drastically when the function
171 * using alloca is called in loop. In GCC present in SPEC2000 inlining
172 * into schedule_block cause it to require 2GB of ram instead of 256MB.
174 * Sorrily this is true with our implementation also.
175 * Moreover, we cannot differentiate between alloca() and VLA yet, so
176 * this disables inlining of functions using VLA (which are completely
180 * - add a flag to the Alloc node for "real" alloca() calls
181 * - add a new Stack-Restore node at the end of a function using
184 *allow_inline = false;
189 * Check if we can inline a given call.
190 * Currently, we cannot inline two cases:
191 * - call with compound arguments
192 * - graphs that take the address of a parameter
194 * check these conditions here
196 static bool can_inline(ir_node *call, ir_graph *called_graph)
198 ir_entity *called = get_irg_entity(called_graph);
199 ir_type *called_type = get_entity_type(called);
200 ir_type *call_type = get_Call_type(call);
201 size_t n_params = get_method_n_params(called_type);
202 size_t n_arguments = get_method_n_params(call_type);
203 size_t n_res = get_method_n_ress(called_type);
204 irg_inline_property prop = get_irg_inline_property(called_graph);
208 if (prop == irg_inline_forbidden)
211 if (n_arguments != n_params) {
212 /* this is a bad feature of C: without a prototype, we can
213 * call a function with less parameters than needed. Currently
214 * we don't support this, although we could use Unknown than. */
217 if (n_res != get_method_n_ress(call_type)) {
221 /* Argh, compiling C has some bad consequences:
222 * It is implementation dependent what happens in that case.
223 * We support inlining, if the bitsize of the types matches AND
224 * the same arithmetic is used. */
225 for (i = 0; i < n_params; ++i) {
226 ir_type *param_tp = get_method_param_type(called_type, i);
227 ir_type *arg_tp = get_method_param_type(call_type, i);
229 if (param_tp != arg_tp) {
230 ir_mode *pmode = get_type_mode(param_tp);
231 ir_mode *amode = get_type_mode(arg_tp);
233 if (pmode == NULL || amode == NULL)
235 if (get_mode_size_bits(pmode) != get_mode_size_bits(amode))
237 if (get_mode_arithmetic(pmode) != get_mode_arithmetic(amode))
239 /* otherwise we can simply "reinterpret" the bits */
242 for (i = 0; i < n_res; ++i) {
243 ir_type *decl_res_tp = get_method_res_type(called_type, i);
244 ir_type *used_res_tp = get_method_res_type(call_type, i);
246 if (decl_res_tp != used_res_tp) {
247 ir_mode *decl_mode = get_type_mode(decl_res_tp);
248 ir_mode *used_mode = get_type_mode(used_res_tp);
249 if (decl_mode == NULL || used_mode == NULL)
251 if (get_mode_size_bits(decl_mode) != get_mode_size_bits(used_mode))
253 if (get_mode_arithmetic(decl_mode) != get_mode_arithmetic(used_mode))
255 /* otherwise we can "reinterpret" the bits */
259 /* check parameters for compound arguments */
260 for (i = 0; i < n_params; ++i) {
261 ir_type *p_type = get_method_param_type(call_type, i);
263 if (is_compound_type(p_type))
267 /* check results for compound arguments */
268 for (i = 0; i < n_res; ++i) {
269 ir_type *r_type = get_method_res_type(call_type, i);
271 if (is_compound_type(r_type))
276 irg_walk_graph(called_graph, find_addr, NULL, &res);
282 exc_handler, /**< There is a handler. */
283 exc_no_handler /**< Exception handling not represented. */
287 * copy all entities on the stack frame on 1 irg to the stackframe of another.
288 * Sets entity links of the old entities to the copies
290 static void copy_frame_entities(ir_graph *from, ir_graph *to)
292 ir_type *from_frame = get_irg_frame_type(from);
293 ir_type *to_frame = get_irg_frame_type(to);
294 size_t n_members = get_class_n_members(from_frame);
296 assert(from_frame != to_frame);
298 for (i = 0; i < n_members; ++i) {
299 ir_entity *old_ent = get_class_member(from_frame, i);
300 ir_entity *new_ent = copy_entity_own(old_ent, to_frame);
301 set_entity_link(old_ent, new_ent);
302 assert (!is_parameter_entity(old_ent));
306 /* Inlines a method at the given call site. */
307 int inline_method(ir_node *call, ir_graph *called_graph)
310 ir_node *post_call, *post_bl;
311 ir_node *in[pn_Start_max+1];
312 ir_node *end, *end_bl, *block;
317 int arity, n_ret, n_exc, n_res, i, j, rem_opt;
318 int irn_arity, n_params;
320 enum exc_mode exc_handling;
325 ir_graph *irg = get_irn_irg(call);
327 /* we cannot inline some types of calls */
328 if (! can_inline(call, called_graph))
331 /* We cannot inline a recursive call. The graph must be copied before
332 * the call the inline_method() using create_irg_copy(). */
333 if (called_graph == irg)
336 ent = get_irg_entity(called_graph);
337 mtp = get_entity_type(ent);
338 ctp = get_Call_type(call);
339 n_params = get_method_n_params(mtp);
340 n_res = get_method_n_ress(mtp);
342 rem = current_ir_graph;
343 current_ir_graph = irg;
345 DB((dbg, LEVEL_1, "Inlining %+F(%+F) into %+F\n", call, called_graph, irg));
347 /* optimizations can cause problems when allocating new nodes */
348 rem_opt = get_opt_optimize();
351 /* Handle graph state */
352 assert(get_irg_phase_state(irg) != phase_building);
353 assert(get_irg_pinned(irg) == op_pin_state_pinned);
354 assert(get_irg_pinned(called_graph) == op_pin_state_pinned);
355 clear_irg_properties(irg, IR_GRAPH_PROPERTY_CONSISTENT_DOMINANCE
356 | IR_GRAPH_PROPERTY_CONSISTENT_ENTITY_USAGE);
357 set_irg_callee_info_state(irg, irg_callee_info_inconsistent);
358 clear_irg_properties(irg, IR_GRAPH_PROPERTY_CONSISTENT_ENTITY_USAGE);
359 edges_deactivate(irg);
361 /* here we know we WILL inline, so inform the statistics */
362 hook_inline(call, called_graph);
364 /* -- Decide how to handle exception control flow: Is there a handler
365 for the Call node, or do we branch directly to End on an exception?
367 0 There is a handler.
368 2 Exception handling not represented in Firm. -- */
370 ir_node *Xproj = NULL;
372 for (proj = (ir_node*)get_irn_link(call); proj != NULL;
373 proj = (ir_node*)get_irn_link(proj)) {
374 long proj_nr = get_Proj_proj(proj);
375 if (proj_nr == pn_Call_X_except) Xproj = proj;
377 exc_handling = Xproj != NULL ? exc_handler : exc_no_handler;
380 /* create the argument tuple */
381 args_in = ALLOCAN(ir_node*, n_params);
383 block = get_nodes_block(call);
384 for (i = n_params - 1; i >= 0; --i) {
385 ir_node *arg = get_Call_param(call, i);
386 ir_type *param_tp = get_method_param_type(mtp, i);
387 ir_mode *mode = get_type_mode(param_tp);
389 if (mode != get_irn_mode(arg)) {
390 arg = new_r_Conv(block, arg, mode);
395 /* the procedure and later replaces the Start node of the called graph.
396 * Post_call is the old Call node and collects the results of the called
397 * graph. Both will end up being a tuple. */
398 post_bl = get_nodes_block(call);
399 /* XxMxPxPxPxT of Start + parameter of Call */
400 in[pn_Start_M] = get_Call_mem(call);
401 in[pn_Start_X_initial_exec] = new_r_Jmp(post_bl);
402 in[pn_Start_P_frame_base] = get_irg_frame(irg);
403 in[pn_Start_T_args] = new_r_Tuple(post_bl, n_params, args_in);
404 pre_call = new_r_Tuple(post_bl, pn_Start_max+1, in);
408 The new block gets the ins of the old block, pre_call and all its
409 predecessors and all Phi nodes. -- */
410 part_block(pre_call);
412 /* increment visited flag for later walk */
413 inc_irg_visited(called_graph);
415 /* link some nodes to nodes in the current graph so instead of copying
416 * the linked nodes will get used.
417 * So the copier will use the created Tuple instead of copying the start
418 * node, similar for singleton nodes like NoMem and Bad.
419 * Note: this will prohibit predecessors to be copied - only do it for
420 * nodes without predecessors */
422 ir_node *start_block;
426 start_block = get_irg_start_block(called_graph);
427 set_new_node(start_block, get_nodes_block(pre_call));
428 mark_irn_visited(start_block);
430 start = get_irg_start(called_graph);
431 set_new_node(start, pre_call);
432 mark_irn_visited(start);
434 nomem = get_irg_no_mem(called_graph);
435 set_new_node(nomem, get_irg_no_mem(irg));
436 mark_irn_visited(nomem);
439 /* entitiy link is used to link entities on old stackframe to the
441 irp_reserve_resources(irp, IRP_RESOURCE_ENTITY_LINK);
443 /* copy entities and nodes */
444 assert(!irn_visited(get_irg_end(called_graph)));
445 copy_frame_entities(called_graph, irg);
446 irg_walk_core(get_irg_end(called_graph), copy_node_inline, set_preds_inline,
449 irp_free_resources(irp, IRP_RESOURCE_ENTITY_LINK);
451 /* -- Merge the end of the inlined procedure with the call site -- */
452 /* We will turn the old Call node into a Tuple with the following
455 0: Phi of all Memories of Return statements.
456 1: Jmp from new Block that merges the control flow from all exception
457 predecessors of the old end block.
458 2: Tuple of all arguments.
459 3: Phi of Exception memories.
460 In case the old Call directly branches to End on an exception we don't
461 need the block merging all exceptions nor the Phi of the exception
465 /* Precompute some values */
466 end_bl = get_new_node(get_irg_end_block(called_graph));
467 end = get_new_node(get_irg_end(called_graph));
468 arity = get_Block_n_cfgpreds(end_bl); /* arity = n_exc + n_ret */
469 n_res = get_method_n_ress(get_Call_type(call));
471 res_pred = XMALLOCN(ir_node*, n_res);
472 cf_pred = XMALLOCN(ir_node*, arity);
474 /* archive keepalives */
475 irn_arity = get_irn_arity(end);
476 for (i = 0; i < irn_arity; i++) {
477 ir_node *ka = get_End_keepalive(end, i);
479 add_End_keepalive(get_irg_end(irg), ka);
482 /* replace Return nodes by Jump nodes */
484 for (i = 0; i < arity; i++) {
486 ret = get_Block_cfgpred(end_bl, i);
487 if (is_Return(ret)) {
488 ir_node *block = get_nodes_block(ret);
489 cf_pred[n_ret] = new_r_Jmp(block);
493 set_irn_in(post_bl, n_ret, cf_pred);
495 /* build a Tuple for all results of the method.
496 * add Phi node if there was more than one Return. */
497 turn_into_tuple(post_call, pn_Call_max+1);
498 /* First the Memory-Phi */
500 for (i = 0; i < arity; i++) {
501 ret = get_Block_cfgpred(end_bl, i);
502 if (is_Return(ret)) {
503 cf_pred[n_mem_phi++] = get_Return_mem(ret);
505 /* memory output for some exceptions is directly connected to End */
507 cf_pred[n_mem_phi++] = new_r_Proj(ret, mode_M, 3);
508 } else if (is_fragile_op(ret)) {
509 /* We rely that all cfops have the memory output at the same position. */
510 cf_pred[n_mem_phi++] = new_r_Proj(ret, mode_M, 0);
511 } else if (is_Raise(ret)) {
512 cf_pred[n_mem_phi++] = new_r_Proj(ret, mode_M, 1);
515 phi = new_r_Phi(post_bl, n_mem_phi, cf_pred, mode_M);
516 set_Tuple_pred(call, pn_Call_M, phi);
517 /* Conserve Phi-list for further inlinings -- but might be optimized */
518 if (get_nodes_block(phi) == post_bl) {
519 set_irn_link(phi, get_irn_link(post_bl));
520 set_irn_link(post_bl, phi);
522 /* Now the real results */
524 ir_node *result_tuple;
525 for (j = 0; j < n_res; j++) {
526 ir_type *res_type = get_method_res_type(ctp, j);
527 ir_mode *res_mode = get_type_mode(res_type);
529 for (i = 0; i < arity; i++) {
530 ret = get_Block_cfgpred(end_bl, i);
531 if (is_Return(ret)) {
532 ir_node *res = get_Return_res(ret, j);
533 if (get_irn_mode(res) != res_mode) {
534 ir_node *block = get_nodes_block(res);
535 res = new_r_Conv(block, res, res_mode);
537 cf_pred[n_ret] = res;
542 phi = new_r_Phi(post_bl, n_ret, cf_pred, res_mode);
544 phi = new_r_Bad(irg, res_mode);
547 /* Conserve Phi-list for further inlinings -- but might be optimized */
548 if (get_nodes_block(phi) == post_bl) {
549 set_Phi_next(phi, get_Block_phis(post_bl));
550 set_Block_phis(post_bl, phi);
553 result_tuple = new_r_Tuple(post_bl, n_res, res_pred);
554 set_Tuple_pred(call, pn_Call_T_result, result_tuple);
556 set_Tuple_pred(call, pn_Call_T_result, new_r_Bad(irg, mode_T));
558 /* handle the regular call */
559 set_Tuple_pred(call, pn_Call_X_regular, new_r_Jmp(post_bl));
561 /* Finally the exception control flow.
562 We have two possible situations:
563 First if the Call branches to an exception handler:
564 We need to add a Phi node to
565 collect the memory containing the exception objects. Further we need
566 to add another block to get a correct representation of this Phi. To
567 this block we add a Jmp that resolves into the X output of the Call
568 when the Call is turned into a tuple.
569 Second: There is no exception edge. Just add all inlined exception
570 branches to the End node.
572 if (exc_handling == exc_handler) {
574 for (i = 0; i < arity; i++) {
576 ret = get_Block_cfgpred(end_bl, i);
577 irn = skip_Proj(ret);
578 if (is_fragile_op(irn) || is_Raise(irn)) {
579 cf_pred[n_exc] = ret;
586 set_Tuple_pred(call, pn_Call_X_except, cf_pred[0]);
588 ir_node *block = new_r_Block(irg, n_exc, cf_pred);
589 set_Tuple_pred(call, pn_Call_X_except, new_r_Jmp(block));
592 set_Tuple_pred(call, pn_Call_X_except, new_r_Bad(irg, mode_X));
595 ir_node *main_end_bl;
596 int main_end_bl_arity;
599 /* assert(exc_handling == 1 || no exceptions. ) */
601 for (i = 0; i < arity; i++) {
602 ir_node *ret = get_Block_cfgpred(end_bl, i);
603 ir_node *irn = skip_Proj(ret);
605 if (is_fragile_op(irn) || is_Raise(irn)) {
606 cf_pred[n_exc] = ret;
610 main_end_bl = get_irg_end_block(irg);
611 main_end_bl_arity = get_irn_arity(main_end_bl);
612 end_preds = XMALLOCN(ir_node*, n_exc + main_end_bl_arity);
614 for (i = 0; i < main_end_bl_arity; ++i)
615 end_preds[i] = get_irn_n(main_end_bl, i);
616 for (i = 0; i < n_exc; ++i)
617 end_preds[main_end_bl_arity + i] = cf_pred[i];
618 set_irn_in(main_end_bl, n_exc + main_end_bl_arity, end_preds);
619 set_Tuple_pred(call, pn_Call_X_except, new_r_Bad(irg, mode_X));
625 /* -- Turn CSE back on. -- */
626 set_optimize(rem_opt);
627 current_ir_graph = rem;
632 /********************************************************************/
633 /* Apply inlining to small methods. */
634 /********************************************************************/
636 static struct obstack temp_obst;
638 /** Represents a possible inlinable call in a graph. */
639 typedef struct call_entry {
640 ir_node *call; /**< The Call node. */
641 ir_graph *callee; /**< The callee IR-graph. */
642 list_head list; /**< List head for linking the next one. */
643 int loop_depth; /**< The loop depth of this call. */
644 int benefice; /**< The calculated benefice of this call. */
645 unsigned local_adr:1; /**< Set if this call gets an address of a local variable. */
646 unsigned all_const:1; /**< Set if this call has only constant parameters. */
650 * environment for inlining small irgs
652 typedef struct inline_env_t {
653 struct obstack obst; /**< An obstack where call_entries are allocated on. */
654 list_head calls; /**< The call entry list. */
658 * Returns the irg called from a Call node. If the irg is not
659 * known, NULL is returned.
661 * @param call the call node
663 static ir_graph *get_call_called_irg(ir_node *call)
667 addr = get_Call_ptr(call);
668 if (is_SymConst_addr_ent(addr)) {
669 ir_entity *ent = get_SymConst_entity(addr);
670 /* we don't know which function gets finally bound to a weak symbol */
671 if (get_entity_linkage(ent) & IR_LINKAGE_WEAK)
674 return get_entity_irg(ent);
681 * Walker: Collect all calls to known graphs inside a graph.
683 static void collect_calls(ir_node *call, void *env)
687 ir_graph *called_irg = get_call_called_irg(call);
689 if (called_irg != NULL) {
690 /* The Call node calls a locally defined method. Remember to inline. */
691 inline_env_t *ienv = (inline_env_t*)env;
692 call_entry *entry = OALLOC(&ienv->obst, call_entry);
694 entry->callee = called_irg;
695 entry->loop_depth = 0;
697 entry->local_adr = 0;
698 entry->all_const = 0;
700 list_add_tail(&entry->list, &ienv->calls);
706 * Inlines all small methods at call sites where the called address comes
707 * from a Const node that references the entity representing the called
709 * The size argument is a rough measure for the code size of the method:
710 * Methods where the obstack containing the firm graph is smaller than
713 void inline_small_irgs(ir_graph *irg, int size)
715 ir_graph *rem = current_ir_graph;
718 current_ir_graph = irg;
719 /* Handle graph state */
720 assert(get_irg_phase_state(irg) != phase_building);
721 free_callee_info(irg);
723 /* Find Call nodes to inline.
724 (We can not inline during a walk of the graph, as inlining the same
725 method several times changes the visited flag of the walked graph:
726 after the first inlining visited of the callee equals visited of
727 the caller. With the next inlining both are increased.) */
728 obstack_init(&env.obst);
729 INIT_LIST_HEAD(&env.calls);
730 irg_walk_graph(irg, NULL, collect_calls, &env);
732 if (! list_empty(&env.calls)) {
733 /* There are calls to inline */
734 ir_reserve_resources(irg, IR_RESOURCE_IRN_LINK|IR_RESOURCE_PHI_LIST);
735 collect_phiprojs(irg);
737 list_for_each_entry(call_entry, entry, &env.calls, list) {
738 ir_graph *callee = entry->callee;
739 irg_inline_property prop = get_irg_inline_property(callee);
741 if (prop == irg_inline_forbidden) {
745 if (prop >= irg_inline_forced ||
746 _obstack_memory_used(callee->obst) - (int)obstack_room(callee->obst) < size) {
747 inline_method(entry->call, callee);
750 ir_free_resources(irg, IR_RESOURCE_IRN_LINK|IR_RESOURCE_PHI_LIST);
752 obstack_free(&env.obst, NULL);
753 current_ir_graph = rem;
756 typedef struct inline_small_irgs_pass_t {
757 ir_graph_pass_t pass;
759 } inline_small_irgs_pass_t;
762 * Wrapper to run inline_small_irgs() as a pass.
764 static int inline_small_irgs_wrapper(ir_graph *irg, void *context)
766 inline_small_irgs_pass_t *pass = (inline_small_irgs_pass_t*)context;
768 inline_small_irgs(irg, pass->size);
772 /* create a pass for inline_small_irgs() */
773 ir_graph_pass_t *inline_small_irgs_pass(const char *name, int size)
775 inline_small_irgs_pass_t *pass = XMALLOCZ(inline_small_irgs_pass_t);
778 return def_graph_pass_constructor(
779 &pass->pass, name ? name : "inline_small_irgs", inline_small_irgs_wrapper);
783 * Environment for inlining irgs.
786 list_head calls; /**< List of of all call nodes in this graph. */
787 unsigned *local_weights; /**< Once allocated, the beneficial weight for transmitting local addresses. */
788 unsigned n_nodes; /**< Number of nodes in graph except Id, Tuple, Proj, Start, End. */
789 unsigned n_blocks; /**< Number of Blocks in graph without Start and End block. */
790 unsigned n_nodes_orig; /**< for statistics */
791 unsigned n_call_nodes; /**< Number of Call nodes in the graph. */
792 unsigned n_call_nodes_orig; /**< for statistics */
793 unsigned n_callers; /**< Number of known graphs that call this graphs. */
794 unsigned n_callers_orig; /**< for statistics */
795 unsigned got_inline:1; /**< Set, if at least one call inside this graph was inlined. */
796 unsigned recursive:1; /**< Set, if this function is self recursive. */
800 * Allocate a new environment for inlining.
802 static inline_irg_env *alloc_inline_irg_env(void)
804 inline_irg_env *env = OALLOC(&temp_obst, inline_irg_env);
805 INIT_LIST_HEAD(&env->calls);
806 env->local_weights = NULL;
807 env->n_nodes = -2; /* do not count count Start, End */
808 env->n_blocks = -2; /* do not count count Start, End Block */
809 env->n_nodes_orig = -2; /* do not count Start, End */
810 env->n_call_nodes = 0;
811 env->n_call_nodes_orig = 0;
813 env->n_callers_orig = 0;
819 typedef struct walker_env {
820 inline_irg_env *x; /**< the inline environment */
821 char ignore_runtime; /**< the ignore runtime flag */
822 char ignore_callers; /**< if set, do change callers data */
826 * post-walker: collect all calls in the inline-environment
827 * of a graph and sum some statistics.
829 static void collect_calls2(ir_node *call, void *ctx)
831 wenv_t *env = (wenv_t*)ctx;
832 inline_irg_env *x = env->x;
833 unsigned code = get_irn_opcode(call);
837 /* count meaningful nodes in irg */
838 if (code != iro_Proj && code != iro_Tuple && code != iro_Sync) {
839 if (code != iro_Block) {
847 if (code != iro_Call) return;
849 /* check, if it's a runtime call */
850 if (env->ignore_runtime) {
851 ir_node *symc = get_Call_ptr(call);
853 if (is_SymConst_addr_ent(symc)) {
854 ir_entity *ent = get_SymConst_entity(symc);
856 if (get_entity_additional_properties(ent) & mtp_property_runtime)
861 /* collect all call nodes */
863 ++x->n_call_nodes_orig;
865 callee = get_call_called_irg(call);
866 if (callee != NULL) {
867 if (! env->ignore_callers) {
868 inline_irg_env *callee_env = (inline_irg_env*)get_irg_link(callee);
869 /* count all static callers */
870 ++callee_env->n_callers;
871 ++callee_env->n_callers_orig;
873 if (callee == current_ir_graph)
876 /* link it in the list of possible inlinable entries */
877 entry = OALLOC(&temp_obst, call_entry);
879 entry->callee = callee;
880 entry->loop_depth = get_irn_loop(get_nodes_block(call))->depth;
882 entry->local_adr = 0;
883 entry->all_const = 0;
885 list_add_tail(&entry->list, &x->calls);
890 * Returns TRUE if the number of callers is 0 in the irg's environment,
891 * hence this irg is a leaf.
893 inline static int is_leaf(ir_graph *irg)
895 inline_irg_env *env = (inline_irg_env*)get_irg_link(irg);
896 return env->n_call_nodes == 0;
900 * Returns TRUE if the number of nodes in the callee is
901 * smaller then size in the irg's environment.
903 inline static int is_smaller(ir_graph *callee, unsigned size)
905 inline_irg_env *env = (inline_irg_env*)get_irg_link(callee);
906 return env->n_nodes < size;
910 * Duplicate a call entry.
912 * @param entry the original entry to duplicate
913 * @param new_call the new call node
914 * @param loop_depth_delta
915 * delta value for the loop depth
917 static call_entry *duplicate_call_entry(const call_entry *entry,
918 ir_node *new_call, int loop_depth_delta)
920 call_entry *nentry = OALLOC(&temp_obst, call_entry);
921 nentry->call = new_call;
922 nentry->callee = entry->callee;
923 nentry->benefice = entry->benefice;
924 nentry->loop_depth = entry->loop_depth + loop_depth_delta;
925 nentry->local_adr = entry->local_adr;
926 nentry->all_const = entry->all_const;
932 * Append all call nodes of the source environment to the nodes of in the destination
935 * @param dst destination environment
936 * @param src source environment
937 * @param loop_depth the loop depth of the call that is replaced by the src list
939 static void append_call_list(inline_irg_env *dst, inline_irg_env *src, int loop_depth)
943 /* Note that the src list points to Call nodes in the inlined graph, but
944 we need Call nodes in our graph. Luckily the inliner leaves this information
945 in the link field. */
946 list_for_each_entry(call_entry, entry, &src->calls, list) {
947 nentry = duplicate_call_entry(entry, (ir_node*)get_irn_link(entry->call), loop_depth);
948 list_add_tail(&nentry->list, &dst->calls);
950 dst->n_call_nodes += src->n_call_nodes;
951 dst->n_nodes += src->n_nodes;
955 * Inlines small leaf methods at call sites where the called address comes
956 * from a Const node that references the entity representing the called
958 * The size argument is a rough measure for the code size of the method:
959 * Methods where the obstack containing the firm graph is smaller than
962 void inline_leaf_functions(unsigned maxsize, unsigned leafsize,
963 unsigned size, int ignore_runtime)
972 pmap_entry *pm_entry;
974 rem = current_ir_graph;
975 obstack_init(&temp_obst);
977 /* a map for the copied graphs, used to inline recursive calls */
978 copied_graphs = pmap_create();
980 /* extend all irgs by a temporary data structure for inlining. */
981 n_irgs = get_irp_n_irgs();
982 for (i = 0; i < n_irgs; ++i)
983 set_irg_link(get_irp_irg(i), alloc_inline_irg_env());
985 /* Pre-compute information in temporary data structure. */
986 wenv.ignore_runtime = ignore_runtime;
987 wenv.ignore_callers = 0;
988 for (i = 0; i < n_irgs; ++i) {
989 ir_graph *irg = get_irp_irg(i);
991 assert(get_irg_phase_state(irg) != phase_building);
992 free_callee_info(irg);
994 assure_irg_properties(irg,
995 IR_GRAPH_PROPERTY_CONSISTENT_LOOPINFO);
996 wenv.x = (inline_irg_env*)get_irg_link(irg);
997 irg_walk_graph(irg, NULL, collect_calls2, &wenv);
998 confirm_irg_properties(irg, IR_GRAPH_PROPERTIES_ALL);
1001 /* -- and now inline. -- */
1003 /* Inline leafs recursively -- we might construct new leafs. */
1007 for (i = 0; i < n_irgs; ++i) {
1009 int phiproj_computed = 0;
1011 current_ir_graph = get_irp_irg(i);
1012 env = (inline_irg_env*)get_irg_link(current_ir_graph);
1014 ir_reserve_resources(current_ir_graph, IR_RESOURCE_IRN_LINK|IR_RESOURCE_PHI_LIST);
1015 list_for_each_entry_safe(call_entry, entry, next, &env->calls, list) {
1017 irg_inline_property prop;
1019 if (env->n_nodes > maxsize)
1023 callee = entry->callee;
1025 prop = get_irg_inline_property(callee);
1026 if (prop == irg_inline_forbidden) {
1030 if (is_leaf(callee) && (
1031 is_smaller(callee, leafsize) || prop >= irg_inline_forced)) {
1032 if (!phiproj_computed) {
1033 phiproj_computed = 1;
1034 collect_phiprojs(current_ir_graph);
1036 did_inline = inline_method(call, callee);
1039 inline_irg_env *callee_env = (inline_irg_env*)get_irg_link(callee);
1041 /* call was inlined, Phi/Projs for current graph must be recomputed */
1042 phiproj_computed = 0;
1044 /* Do some statistics */
1045 env->got_inline = 1;
1046 --env->n_call_nodes;
1047 env->n_nodes += callee_env->n_nodes;
1048 --callee_env->n_callers;
1050 /* remove this call from the list */
1051 list_del(&entry->list);
1056 ir_free_resources(current_ir_graph, IR_RESOURCE_IRN_LINK|IR_RESOURCE_PHI_LIST);
1058 } while (did_inline);
1060 /* inline other small functions. */
1061 for (i = 0; i < n_irgs; ++i) {
1063 int phiproj_computed = 0;
1065 current_ir_graph = get_irp_irg(i);
1066 env = (inline_irg_env*)get_irg_link(current_ir_graph);
1068 ir_reserve_resources(current_ir_graph, IR_RESOURCE_IRN_LINK|IR_RESOURCE_PHI_LIST);
1070 /* note that the list of possible calls is updated during the process */
1071 list_for_each_entry_safe(call_entry, entry, next, &env->calls, list) {
1072 irg_inline_property prop;
1077 callee = entry->callee;
1079 prop = get_irg_inline_property(callee);
1080 if (prop == irg_inline_forbidden) {
1084 calleee = pmap_get(ir_graph, copied_graphs, callee);
1085 if (calleee != NULL) {
1087 * Remap callee if we have a copy.
1088 * FIXME: Should we do this only for recursive Calls ?
1093 if (prop >= irg_inline_forced ||
1094 (is_smaller(callee, size) && env->n_nodes < maxsize) /* small function */) {
1095 if (current_ir_graph == callee) {
1097 * Recursive call: we cannot directly inline because we cannot walk
1098 * the graph and change it. So we have to make a copy of the graph
1102 inline_irg_env *callee_env;
1105 ir_free_resources(current_ir_graph, IR_RESOURCE_IRN_LINK|IR_RESOURCE_PHI_LIST);
1108 * No copy yet, create one.
1109 * Note that recursive methods are never leafs, so it is sufficient
1110 * to test this condition here.
1112 copy = create_irg_copy(callee);
1114 /* create_irg_copy() destroys the Proj links, recompute them */
1115 phiproj_computed = 0;
1117 ir_reserve_resources(current_ir_graph, IR_RESOURCE_IRN_LINK|IR_RESOURCE_PHI_LIST);
1119 /* allocate new environment */
1120 callee_env = alloc_inline_irg_env();
1121 set_irg_link(copy, callee_env);
1123 assure_irg_properties(copy,
1124 IR_GRAPH_PROPERTY_CONSISTENT_LOOPINFO);
1125 wenv.x = callee_env;
1126 wenv.ignore_callers = 1;
1127 irg_walk_graph(copy, NULL, collect_calls2, &wenv);
1130 * Enter the entity of the original graph. This is needed
1131 * for inline_method(). However, note that ent->irg still points
1132 * to callee, NOT to copy.
1134 set_irg_entity(copy, get_irg_entity(callee));
1136 pmap_insert(copied_graphs, callee, copy);
1139 /* we have only one caller: the original graph */
1140 callee_env->n_callers = 1;
1141 callee_env->n_callers_orig = 1;
1143 if (! phiproj_computed) {
1144 phiproj_computed = 1;
1145 collect_phiprojs(current_ir_graph);
1147 did_inline = inline_method(call, callee);
1149 inline_irg_env *callee_env = (inline_irg_env *)get_irg_link(callee);
1151 /* call was inlined, Phi/Projs for current graph must be recomputed */
1152 phiproj_computed = 0;
1154 /* callee was inline. Append its call list. */
1155 env->got_inline = 1;
1156 --env->n_call_nodes;
1157 append_call_list(env, callee_env, entry->loop_depth);
1158 --callee_env->n_callers;
1160 /* after we have inlined callee, all called methods inside callee
1161 are now called once more */
1162 list_for_each_entry(call_entry, centry, &callee_env->calls, list) {
1163 inline_irg_env *penv = (inline_irg_env*)get_irg_link(centry->callee);
1167 /* remove this call from the list */
1168 list_del(&entry->list);
1173 ir_free_resources(current_ir_graph, IR_RESOURCE_IRN_LINK|IR_RESOURCE_PHI_LIST);
1176 for (i = 0; i < n_irgs; ++i) {
1177 irg = get_irp_irg(i);
1178 env = (inline_irg_env*)get_irg_link(irg);
1180 if (env->got_inline) {
1181 optimize_graph_df(irg);
1184 if (env->got_inline || (env->n_callers_orig != env->n_callers)) {
1185 DB((dbg, LEVEL_1, "Nodes:%3d ->%3d, calls:%3d ->%3d, callers:%3d ->%3d, -- %s\n",
1186 env->n_nodes_orig, env->n_nodes, env->n_call_nodes_orig, env->n_call_nodes,
1187 env->n_callers_orig, env->n_callers,
1188 get_entity_name(get_irg_entity(irg))));
1192 /* kill the copied graphs: we don't need them anymore */
1193 foreach_pmap(copied_graphs, pm_entry) {
1194 ir_graph *copy = (ir_graph*)pm_entry->value;
1196 /* reset the entity, otherwise it will be deleted in the next step ... */
1197 set_irg_entity(copy, NULL);
1198 free_ir_graph(copy);
1200 pmap_destroy(copied_graphs);
1202 obstack_free(&temp_obst, NULL);
1203 current_ir_graph = rem;
1206 typedef struct inline_leaf_functions_pass_t {
1207 ir_prog_pass_t pass;
1212 } inline_leaf_functions_pass_t;
1215 * Wrapper to run inline_leaf_functions() as a ir_prog pass.
1217 static int inline_leaf_functions_wrapper(ir_prog *irp, void *context)
1219 inline_leaf_functions_pass_t *pass = (inline_leaf_functions_pass_t*)context;
1222 inline_leaf_functions(
1223 pass->maxsize, pass->leafsize,
1224 pass->size, pass->ignore_runtime);
1228 /* create a pass for inline_leaf_functions() */
1229 ir_prog_pass_t *inline_leaf_functions_pass(
1230 const char *name, unsigned maxsize, unsigned leafsize,
1231 unsigned size, int ignore_runtime)
1233 inline_leaf_functions_pass_t *pass = XMALLOCZ(inline_leaf_functions_pass_t);
1235 pass->maxsize = maxsize;
1236 pass->leafsize = leafsize;
1238 pass->ignore_runtime = ignore_runtime;
1240 return def_prog_pass_constructor(
1242 name ? name : "inline_leaf_functions",
1243 inline_leaf_functions_wrapper);
1247 * Calculate the parameter weights for transmitting the address of a local variable.
1249 static unsigned calc_method_local_weight(ir_node *arg)
1252 unsigned v, weight = 0;
1254 for (unsigned i = get_irn_n_outs(arg); i-- > 0; ) {
1255 ir_node *succ = get_irn_out(arg, i);
1257 switch (get_irn_opcode(succ)) {
1260 /* Loads and Store can be removed */
1264 /* check if all args are constant */
1265 for (j = get_Sel_n_indexs(succ) - 1; j >= 0; --j) {
1266 ir_node *idx = get_Sel_index(succ, j);
1267 if (! is_Const(idx))
1270 /* Check users on this Sel. Note: if a 0 is returned here, there was
1271 some unsupported node. */
1272 v = calc_method_local_weight(succ);
1275 /* we can kill one Sel with constant indexes, this is cheap */
1279 /* when looking backward we might find Id nodes */
1280 weight += calc_method_local_weight(succ);
1283 /* unoptimized tuple */
1284 for (j = get_Tuple_n_preds(succ) - 1; j >= 0; --j) {
1285 ir_node *pred = get_Tuple_pred(succ, j);
1287 /* look for Proj(j) */
1288 for (unsigned k = get_irn_n_outs(succ); k-- > 0; ) {
1289 ir_node *succ_succ = get_irn_out(succ, k);
1290 if (is_Proj(succ_succ)) {
1291 if (get_Proj_proj(succ_succ) == j) {
1293 weight += calc_method_local_weight(succ_succ);
1296 /* this should NOT happen */
1304 /* any other node: unsupported yet or bad. */
1312 * Calculate the parameter weights for transmitting the address of a local variable.
1314 static void analyze_irg_local_weights(inline_irg_env *env, ir_graph *irg)
1316 ir_entity *ent = get_irg_entity(irg);
1320 ir_node *irg_args, *arg;
1322 mtp = get_entity_type(ent);
1323 nparams = get_method_n_params(mtp);
1325 /* allocate a new array. currently used as 'analysed' flag */
1326 env->local_weights = NEW_ARR_D(unsigned, &temp_obst, nparams);
1328 /* If the method haven't parameters we have nothing to do. */
1332 assure_irg_outs(irg);
1333 irg_args = get_irg_args(irg);
1334 for (unsigned i = get_irn_n_outs(irg_args); i-- > 0; ) {
1335 arg = get_irn_out(irg_args, i);
1336 proj_nr = get_Proj_proj(arg);
1337 env->local_weights[proj_nr] = calc_method_local_weight(arg);
1342 * Calculate the benefice for transmitting an local variable address.
1343 * After inlining, the local variable might be transformed into a
1344 * SSA variable by scalar_replacement().
1346 static unsigned get_method_local_adress_weight(ir_graph *callee, size_t pos)
1348 inline_irg_env *env = (inline_irg_env*)get_irg_link(callee);
1350 if (env->local_weights == NULL)
1351 analyze_irg_local_weights(env, callee);
1353 if (pos < ARR_LEN(env->local_weights))
1354 return env->local_weights[pos];
1359 * Calculate a benefice value for inlining the given call.
1361 * @param call the call node we have to inspect
1362 * @param callee the called graph
1364 static int calc_inline_benefice(call_entry *entry, ir_graph *callee)
1366 ir_node *call = entry->call;
1367 ir_entity *ent = get_irg_entity(callee);
1368 ir_type *callee_frame;
1369 size_t i, n_members, n_params;
1375 irg_inline_property prop;
1377 inline_irg_env *callee_env;
1379 prop = get_irg_inline_property(callee);
1380 if (prop == irg_inline_forbidden) {
1381 DB((dbg, LEVEL_2, "In %+F Call to %+F: inlining forbidden\n",
1383 return entry->benefice = INT_MIN;
1386 callee_frame = get_irg_frame_type(callee);
1387 n_members = get_class_n_members(callee_frame);
1388 for (i = 0; i < n_members; ++i) {
1389 ir_entity *frame_ent = get_class_member(callee_frame, i);
1390 if (is_parameter_entity(frame_ent)) {
1391 // TODO inliner should handle parameter entities by inserting Store operations
1392 DB((dbg, LEVEL_2, "In %+F Call to %+F: inlining forbidden due to parameter entity\n", call, callee));
1393 set_irg_inline_property(callee, irg_inline_forbidden);
1394 return entry->benefice = INT_MIN;
1398 if (get_irg_additional_properties(callee) & mtp_property_noreturn) {
1399 DB((dbg, LEVEL_2, "In %+F Call to %+F: not inlining noreturn or weak\n",
1401 return entry->benefice = INT_MIN;
1404 /* costs for every passed parameter */
1405 n_params = get_Call_n_params(call);
1406 mtp = get_entity_type(ent);
1407 cc = get_method_calling_convention(mtp);
1408 if (cc & cc_reg_param) {
1409 /* register parameter, smaller costs for register parameters */
1410 size_t max_regs = cc & ~cc_bits;
1412 if (max_regs < n_params)
1413 weight += max_regs * 2 + (n_params - max_regs) * 5;
1415 weight += n_params * 2;
1417 /* parameters are passed an stack */
1418 weight += 5 * n_params;
1421 /* constant parameters improve the benefice */
1422 frame_ptr = get_irg_frame(current_ir_graph);
1424 for (i = 0; i < n_params; ++i) {
1425 ir_node *param = get_Call_param(call, i);
1427 if (is_Const(param)) {
1428 weight += get_method_param_weight(ent, i);
1431 if (is_SymConst(param))
1432 weight += get_method_param_weight(ent, i);
1433 else if (is_Sel(param) && get_Sel_ptr(param) == frame_ptr) {
1435 * An address of a local variable is transmitted. After
1436 * inlining, scalar_replacement might be able to remove the
1437 * local variable, so honor this.
1439 v = get_method_local_adress_weight(callee, i);
1442 entry->local_adr = 1;
1446 entry->all_const = all_const;
1448 callee_env = (inline_irg_env*)get_irg_link(callee);
1449 if (callee_env->n_callers == 1 &&
1450 callee != current_ir_graph &&
1451 !entity_is_externally_visible(ent)) {
1455 /* give a bonus for functions with one block */
1456 if (callee_env->n_blocks == 1)
1457 weight = weight * 3 / 2;
1459 /* and one for small non-recursive functions: we want them to be inlined in mostly every case */
1460 if (callee_env->n_nodes < 30 && !callee_env->recursive)
1463 /* and finally for leafs: they do not increase the register pressure
1464 because of callee safe registers */
1465 if (callee_env->n_call_nodes == 0)
1468 /** it's important to inline inner loops first */
1469 if (entry->loop_depth > 30)
1470 weight += 30 * 1024;
1472 weight += entry->loop_depth * 1024;
1475 * All arguments constant is probably a good sign, give an extra bonus
1480 return entry->benefice = weight;
1483 typedef struct walk_env_t {
1489 * Callgraph walker, collect all visited graphs.
1491 static void callgraph_walker(ir_graph *irg, void *data)
1493 walk_env_t *env = (walk_env_t *)data;
1494 env->irgs[env->last_irg++] = irg;
1498 * Creates an inline order for all graphs.
1500 * @return the list of graphs.
1502 static ir_graph **create_irg_list(void)
1504 ir_entity **free_methods;
1505 size_t n_irgs = get_irp_n_irgs();
1508 cgana(&free_methods);
1509 xfree(free_methods);
1511 compute_callgraph();
1513 env.irgs = XMALLOCNZ(ir_graph*, n_irgs);
1516 callgraph_walk(NULL, callgraph_walker, &env);
1517 assert(n_irgs == env.last_irg);
1525 * Push a call onto the priority list if its benefice is big enough.
1527 * @param pqueue the priority queue of calls
1528 * @param call the call entry
1529 * @param inlien_threshold
1530 * the threshold value
1532 static void maybe_push_call(pqueue_t *pqueue, call_entry *call,
1533 int inline_threshold)
1535 ir_graph *callee = call->callee;
1536 irg_inline_property prop = get_irg_inline_property(callee);
1537 int benefice = calc_inline_benefice(call, callee);
1539 DB((dbg, LEVEL_2, "In %+F Call %+F to %+F has benefice %d\n",
1540 get_irn_irg(call->call), call->call, callee, benefice));
1542 if (prop < irg_inline_forced && benefice < inline_threshold) {
1546 pqueue_put(pqueue, call, benefice);
1550 * Try to inline calls into a graph.
1552 * @param irg the graph into which we inline
1553 * @param maxsize do NOT inline if the size of irg gets
1554 * bigger than this amount
1555 * @param inline_threshold
1556 * threshold value for inline decision
1557 * @param copied_graphs
1558 * map containing copied of recursive graphs
1560 static void inline_into(ir_graph *irg, unsigned maxsize,
1561 int inline_threshold, pmap *copied_graphs)
1563 int phiproj_computed = 0;
1564 inline_irg_env *env = (inline_irg_env*)get_irg_link(irg);
1568 if (env->n_call_nodes == 0)
1571 if (env->n_nodes > maxsize) {
1572 DB((dbg, LEVEL_2, "%+F: too big (%d)\n", irg, env->n_nodes));
1576 current_ir_graph = irg;
1577 ir_reserve_resources(irg, IR_RESOURCE_IRN_LINK|IR_RESOURCE_PHI_LIST);
1579 /* put irgs into the pqueue */
1580 pqueue = new_pqueue();
1582 list_for_each_entry(call_entry, curr_call, &env->calls, list) {
1583 assert(is_Call(curr_call->call));
1584 maybe_push_call(pqueue, curr_call, inline_threshold);
1587 /* note that the list of possible calls is updated during the process */
1588 while (!pqueue_empty(pqueue)) {
1590 call_entry *curr_call = (call_entry*)pqueue_pop_front(pqueue);
1591 ir_graph *callee = curr_call->callee;
1592 ir_node *call_node = curr_call->call;
1593 inline_irg_env *callee_env = (inline_irg_env*)get_irg_link(callee);
1594 irg_inline_property prop = get_irg_inline_property(callee);
1598 if ((prop < irg_inline_forced) && env->n_nodes + callee_env->n_nodes > maxsize) {
1599 DB((dbg, LEVEL_2, "%+F: too big (%d) + %+F (%d)\n", irg,
1600 env->n_nodes, callee, callee_env->n_nodes));
1604 calleee = pmap_get(ir_graph, copied_graphs, callee);
1605 if (calleee != NULL) {
1606 int benefice = curr_call->benefice;
1608 * Reduce the weight for recursive function IFF not all arguments are const.
1609 * inlining recursive functions is rarely good.
1611 if (!curr_call->all_const)
1613 if (benefice < inline_threshold)
1617 * Remap callee if we have a copy.
1620 callee_env = (inline_irg_env*)get_irg_link(callee);
1623 if (current_ir_graph == callee) {
1625 * Recursive call: we cannot directly inline because we cannot
1626 * walk the graph and change it. So we have to make a copy of
1629 int benefice = curr_call->benefice;
1633 * Reduce the weight for recursive function IFF not all arguments are const.
1634 * inlining recursive functions is rarely good.
1636 if (!curr_call->all_const)
1638 if (benefice < inline_threshold)
1641 ir_free_resources(irg, IR_RESOURCE_IRN_LINK|IR_RESOURCE_PHI_LIST);
1644 * No copy yet, create one.
1645 * Note that recursive methods are never leafs, so it is
1646 * sufficient to test this condition here.
1648 copy = create_irg_copy(callee);
1650 /* create_irg_copy() destroys the Proj links, recompute them */
1651 phiproj_computed = 0;
1653 ir_reserve_resources(irg, IR_RESOURCE_IRN_LINK|IR_RESOURCE_PHI_LIST);
1655 /* allocate a new environment */
1656 callee_env = alloc_inline_irg_env();
1657 set_irg_link(copy, callee_env);
1659 assure_irg_properties(copy, IR_GRAPH_PROPERTY_CONSISTENT_LOOPINFO);
1660 memset(&wenv, 0, sizeof(wenv));
1661 wenv.x = callee_env;
1662 wenv.ignore_callers = 1;
1663 irg_walk_graph(copy, NULL, collect_calls2, &wenv);
1666 * Enter the entity of the original graph. This is needed
1667 * for inline_method(). However, note that ent->irg still points
1668 * to callee, NOT to copy.
1670 set_irg_entity(copy, get_irg_entity(callee));
1672 pmap_insert(copied_graphs, callee, copy);
1675 /* we have only one caller: the original graph */
1676 callee_env->n_callers = 1;
1677 callee_env->n_callers_orig = 1;
1679 if (! phiproj_computed) {
1680 phiproj_computed = 1;
1681 collect_phiprojs(current_ir_graph);
1683 did_inline = inline_method(call_node, callee);
1687 /* call was inlined, Phi/Projs for current graph must be recomputed */
1688 phiproj_computed = 0;
1690 /* remove it from the caller list */
1691 list_del(&curr_call->list);
1693 /* callee was inline. Append its call list. */
1694 env->got_inline = 1;
1695 --env->n_call_nodes;
1697 /* we just generate a bunch of new calls */
1698 loop_depth = curr_call->loop_depth;
1699 list_for_each_entry(call_entry, centry, &callee_env->calls, list) {
1700 inline_irg_env *penv = (inline_irg_env*)get_irg_link(centry->callee);
1702 call_entry *new_entry;
1704 /* after we have inlined callee, all called methods inside
1705 * callee are now called once more */
1708 /* Note that the src list points to Call nodes in the inlined graph,
1709 * but we need Call nodes in our graph. Luckily the inliner leaves
1710 * this information in the link field. */
1711 new_call = (ir_node*)get_irn_link(centry->call);
1712 if (get_irn_irg(new_call) != irg) {
1713 /* centry->call has not been copied, which means it is dead.
1714 * This might happen during inlining, if a const function,
1715 * which cannot be inlined is only used as an unused argument
1716 * of another function, which is inlined. */
1719 assert(is_Call(new_call));
1721 new_entry = duplicate_call_entry(centry, new_call, loop_depth);
1722 list_add_tail(&new_entry->list, &env->calls);
1723 maybe_push_call(pqueue, new_entry, inline_threshold);
1726 env->n_call_nodes += callee_env->n_call_nodes;
1727 env->n_nodes += callee_env->n_nodes;
1728 --callee_env->n_callers;
1730 ir_free_resources(irg, IR_RESOURCE_IRN_LINK|IR_RESOURCE_PHI_LIST);
1735 * Heuristic inliner. Calculates a benefice value for every call and inlines
1736 * those calls with a value higher than the threshold.
1738 void inline_functions(unsigned maxsize, int inline_threshold,
1739 opt_ptr after_inline_opt)
1741 inline_irg_env *env;
1745 pmap *copied_graphs;
1746 pmap_entry *pm_entry;
1749 rem = current_ir_graph;
1750 obstack_init(&temp_obst);
1752 irgs = create_irg_list();
1754 /* a map for the copied graphs, used to inline recursive calls */
1755 copied_graphs = pmap_create();
1757 /* extend all irgs by a temporary data structure for inlining. */
1758 n_irgs = get_irp_n_irgs();
1759 for (i = 0; i < n_irgs; ++i)
1760 set_irg_link(irgs[i], alloc_inline_irg_env());
1762 /* Pre-compute information in temporary data structure. */
1763 wenv.ignore_runtime = 0;
1764 wenv.ignore_callers = 0;
1765 for (i = 0; i < n_irgs; ++i) {
1766 ir_graph *irg = irgs[i];
1768 free_callee_info(irg);
1770 wenv.x = (inline_irg_env*)get_irg_link(irg);
1771 assure_loopinfo(irg);
1772 irg_walk_graph(irg, NULL, collect_calls2, &wenv);
1775 /* -- and now inline. -- */
1776 for (i = 0; i < n_irgs; ++i) {
1777 ir_graph *irg = irgs[i];
1779 inline_into(irg, maxsize, inline_threshold, copied_graphs);
1782 for (i = 0; i < n_irgs; ++i) {
1783 ir_graph *irg = irgs[i];
1785 env = (inline_irg_env*)get_irg_link(irg);
1786 if (env->got_inline && after_inline_opt != NULL) {
1787 /* this irg got calls inlined: optimize it */
1788 after_inline_opt(irg);
1790 if (env->got_inline || (env->n_callers_orig != env->n_callers)) {
1791 DB((dbg, LEVEL_1, "Nodes:%3d ->%3d, calls:%3d ->%3d, callers:%3d ->%3d, -- %s\n",
1792 env->n_nodes_orig, env->n_nodes, env->n_call_nodes_orig, env->n_call_nodes,
1793 env->n_callers_orig, env->n_callers,
1794 get_entity_name(get_irg_entity(irg))));
1798 /* kill the copied graphs: we don't need them anymore */
1799 foreach_pmap(copied_graphs, pm_entry) {
1800 ir_graph *copy = (ir_graph*)pm_entry->value;
1802 /* reset the entity, otherwise it will be deleted in the next step ... */
1803 set_irg_entity(copy, NULL);
1804 free_ir_graph(copy);
1806 pmap_destroy(copied_graphs);
1810 obstack_free(&temp_obst, NULL);
1811 current_ir_graph = rem;
1814 typedef struct inline_functions_pass_t {
1815 ir_prog_pass_t pass;
1817 int inline_threshold;
1818 opt_ptr after_inline_opt;
1819 } inline_functions_pass_t;
1822 * Wrapper to run inline_functions() as a ir_prog pass.
1824 static int inline_functions_wrapper(ir_prog *irp, void *context)
1826 inline_functions_pass_t *pass = (inline_functions_pass_t*)context;
1829 inline_functions(pass->maxsize, pass->inline_threshold,
1830 pass->after_inline_opt);
1834 /* create a ir_prog pass for inline_functions */
1835 ir_prog_pass_t *inline_functions_pass(
1836 const char *name, unsigned maxsize, int inline_threshold,
1837 opt_ptr after_inline_opt)
1839 inline_functions_pass_t *pass = XMALLOCZ(inline_functions_pass_t);
1841 pass->maxsize = maxsize;
1842 pass->inline_threshold = inline_threshold;
1843 pass->after_inline_opt = after_inline_opt;
1845 return def_prog_pass_constructor(
1846 &pass->pass, name ? name : "inline_functions",
1847 inline_functions_wrapper);
1850 void firm_init_inline(void)
1852 FIRM_DBG_REGISTER(dbg, "firm.opt.inline");