2 * Copyright (C) 1995-2008 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"
53 #include "opt_inline_t.h"
58 #include "analyze_irg_args.h"
59 #include "iredges_t.h"
63 #include "iropt_dbg.h"
65 DEBUG_ONLY(static firm_dbg_module_t *dbg;)
67 /*------------------------------------------------------------------*/
68 /* Routines for dead node elimination / copying garbage collection */
70 /*------------------------------------------------------------------*/
73 * Remember the new node in the old node by using a field all nodes have.
75 #define set_new_node(oldn, newn) set_irn_link(oldn, newn)
78 * Get this new node, before the old node is forgotten.
80 #define get_new_node(oldn) get_irn_link(oldn)
83 * Check if a new node was set.
85 #define has_new_node(n) (get_new_node(n) != NULL)
88 * We use the block_visited flag to mark that we have computed the
89 * number of useful predecessors for this block.
90 * Further we encode the new arity in this flag in the old blocks.
91 * Remembering the arity is useful, as it saves a lot of pointer
92 * accesses. This function is called for all Phi and Block nodes
96 compute_new_arity(ir_node *b) {
97 int i, res, irn_arity;
100 irg_v = get_irg_block_visited(current_ir_graph);
101 block_v = get_Block_block_visited(b);
102 if (block_v >= irg_v) {
103 /* we computed the number of preds for this block and saved it in the
105 return block_v - irg_v;
107 /* compute the number of good predecessors */
108 res = irn_arity = get_irn_arity(b);
109 for (i = 0; i < irn_arity; i++)
110 if (is_Bad(get_irn_n(b, i))) res--;
111 /* save it in the flag. */
112 set_Block_block_visited(b, irg_v + res);
118 * Copies the node to the new obstack. The Ins of the new node point to
119 * the predecessors on the old obstack. For block/phi nodes not all
120 * predecessors might be copied. n->link points to the new node.
121 * For Phi and Block nodes the function allocates in-arrays with an arity
122 * only for useful predecessors. The arity is determined by counting
123 * the non-bad predecessors of the block.
125 * @param n The node to be copied
126 * @param env if non-NULL, the node number attribute will be copied to the new node
128 * Note: Also used for loop unrolling.
130 static void copy_node(ir_node *n, void *env) {
133 ir_op *op = get_irn_op(n);
137 /* node copied already */
139 } else if (op == op_Block) {
141 new_arity = compute_new_arity(n);
142 n->attr.block.graph_arr = NULL;
144 block = get_nodes_block(n);
146 new_arity = compute_new_arity(block);
148 new_arity = get_irn_arity(n);
151 nn = new_ir_node(get_irn_dbg_info(n),
158 /* Copy the attributes. These might point to additional data. If this
159 was allocated on the old obstack the pointers now are dangling. This
160 frees e.g. the memory of the graph_arr allocated in new_immBlock. */
161 if (op == op_Block) {
162 /* we cannot allow blocks WITHOUT macroblock input */
163 set_Block_MacroBlock(nn, get_Block_MacroBlock(n));
165 copy_node_attr(n, nn);
168 /* for easier debugging, we want to copy the node numbers too */
169 nn->node_nr = n->node_nr;
173 hook_dead_node_elim_subst(current_ir_graph, n, nn);
177 * Copies new predecessors of old node to new node remembered in link.
178 * Spare the Bad predecessors of Phi and Block nodes.
180 static void copy_preds(ir_node *n, void *env) {
185 nn = get_new_node(n);
188 /* copy the macro block header */
189 ir_node *mbh = get_Block_MacroBlock(n);
192 /* this block is a macroblock header */
193 set_Block_MacroBlock(nn, nn);
195 /* get the macro block header */
196 ir_node *nmbh = get_new_node(mbh);
197 assert(nmbh != NULL);
198 set_Block_MacroBlock(nn, nmbh);
201 /* Don't copy Bad nodes. */
203 irn_arity = get_irn_arity(n);
204 for (i = 0; i < irn_arity; i++) {
205 if (! is_Bad(get_irn_n(n, i))) {
206 ir_node *pred = get_irn_n(n, i);
207 set_irn_n(nn, j, get_new_node(pred));
211 /* repair the block visited flag from above misuse. Repair it in both
212 graphs so that the old one can still be used. */
213 set_Block_block_visited(nn, 0);
214 set_Block_block_visited(n, 0);
215 /* Local optimization could not merge two subsequent blocks if
216 in array contained Bads. Now it's possible.
217 We don't call optimize_in_place as it requires
218 that the fields in ir_graph are set properly. */
219 if (!has_Block_label(nn) &&
220 get_opt_control_flow_straightening() &&
221 get_Block_n_cfgpreds(nn) == 1 &&
222 is_Jmp(get_Block_cfgpred(nn, 0))) {
223 ir_node *old = get_nodes_block(get_Block_cfgpred(nn, 0));
225 /* Jmp jumps into the block it is in -- deal self cycle. */
226 assert(is_Bad(get_new_node(get_irg_bad(current_ir_graph))));
227 exchange(nn, get_new_node(get_irg_bad(current_ir_graph)));
232 } else if (is_Phi(n) && get_irn_arity(n) > 0) {
233 /* Don't copy node if corresponding predecessor in block is Bad.
234 The Block itself should not be Bad. */
235 block = get_nodes_block(n);
236 set_nodes_block(nn, get_new_node(block));
238 irn_arity = get_irn_arity(n);
239 for (i = 0; i < irn_arity; i++) {
240 if (! is_Bad(get_irn_n(block, i))) {
241 ir_node *pred = get_irn_n(n, i);
242 set_irn_n(nn, j, get_new_node(pred));
243 /*if (is_backedge(n, i)) set_backedge(nn, j);*/
247 /* If the pre walker reached this Phi after the post walker visited the
248 block block_visited is > 0. */
249 set_Block_block_visited(get_nodes_block(n), 0);
250 /* Compacting the Phi's ins might generate Phis with only one
252 if (get_irn_arity(nn) == 1)
253 exchange(nn, get_irn_n(nn, 0));
255 irn_arity = get_irn_arity(n);
256 for (i = -1; i < irn_arity; i++)
257 set_irn_n(nn, i, get_new_node(get_irn_n(n, i)));
259 /* Now the new node is complete. We can add it to the hash table for CSE.
260 @@@ inlining aborts if we identify End. Why? */
262 add_identities(current_ir_graph->value_table, nn);
266 * Copies the graph recursively, compacts the keep-alives of the end node.
268 * @param irg the graph to be copied
269 * @param copy_node_nr If non-zero, the node number will be copied
271 static void copy_graph(ir_graph *irg, int copy_node_nr) {
272 ir_node *oe, *ne, *ob, *nb, *om, *nm; /* old end, new end, old bad, new bad, old NoMem, new NoMem */
273 ir_node *ka; /* keep alive */
277 /* Some nodes must be copied by hand, sigh */
278 vfl = get_irg_visited(irg);
279 set_irg_visited(irg, vfl + 1);
281 oe = get_irg_end(irg);
282 mark_irn_visited(oe);
283 /* copy the end node by hand, allocate dynamic in array! */
284 ne = new_ir_node(get_irn_dbg_info(oe),
291 /* Copy the attributes. Well, there might be some in the future... */
292 copy_node_attr(oe, ne);
293 set_new_node(oe, ne);
295 /* copy the Bad node */
296 ob = get_irg_bad(irg);
297 mark_irn_visited(ob);
298 nb = new_ir_node(get_irn_dbg_info(ob),
305 copy_node_attr(ob, nb);
306 set_new_node(ob, nb);
308 /* copy the NoMem node */
309 om = get_irg_no_mem(irg);
310 mark_irn_visited(om);
311 nm = new_ir_node(get_irn_dbg_info(om),
318 copy_node_attr(om, nm);
319 set_new_node(om, nm);
321 /* copy the live nodes */
322 set_irg_visited(irg, vfl);
323 irg_walk(get_nodes_block(oe), copy_node, copy_preds, INT_TO_PTR(copy_node_nr));
325 /* Note: from yet, the visited flag of the graph is equal to vfl + 1 */
327 /* visit the anchors as well */
328 for (i = get_irg_n_anchors(irg) - 1; i >= 0; --i) {
329 ir_node *n = get_irg_anchor(irg, i);
331 if (n && (get_irn_visited(n) <= vfl)) {
332 set_irg_visited(irg, vfl);
333 irg_walk(n, copy_node, copy_preds, INT_TO_PTR(copy_node_nr));
337 /* copy_preds for the end node ... */
338 set_nodes_block(ne, get_new_node(get_nodes_block(oe)));
340 /*- ... and now the keep alives. -*/
341 /* First pick the not marked block nodes and walk them. We must pick these
342 first as else we will oversee blocks reachable from Phis. */
343 irn_arity = get_End_n_keepalives(oe);
344 for (i = 0; i < irn_arity; i++) {
345 ka = get_End_keepalive(oe, i);
347 if (get_irn_visited(ka) <= vfl) {
348 /* We must keep the block alive and copy everything reachable */
349 set_irg_visited(irg, vfl);
350 irg_walk(ka, copy_node, copy_preds, INT_TO_PTR(copy_node_nr));
352 add_End_keepalive(ne, get_new_node(ka));
356 /* Now pick other nodes. Here we will keep all! */
357 irn_arity = get_End_n_keepalives(oe);
358 for (i = 0; i < irn_arity; i++) {
359 ka = get_End_keepalive(oe, i);
361 if (get_irn_visited(ka) <= vfl) {
362 /* We didn't copy the node yet. */
363 set_irg_visited(irg, vfl);
364 irg_walk(ka, copy_node, copy_preds, INT_TO_PTR(copy_node_nr));
366 add_End_keepalive(ne, get_new_node(ka));
370 /* start block sometimes only reached after keep alives */
371 set_nodes_block(nb, get_new_node(get_nodes_block(ob)));
372 set_nodes_block(nm, get_new_node(get_nodes_block(om)));
376 * Copies the graph reachable from current_ir_graph->end to the obstack
377 * in current_ir_graph and fixes the environment.
378 * Then fixes the fields in current_ir_graph containing nodes of the
381 * @param copy_node_nr If non-zero, the node number will be copied
384 copy_graph_env(int copy_node_nr) {
385 ir_graph *irg = current_ir_graph;
386 ir_node *old_end, *new_anchor;
389 /* remove end_except and end_reg nodes */
390 old_end = get_irg_end(irg);
391 set_irg_end_except (irg, old_end);
392 set_irg_end_reg (irg, old_end);
394 /* Not all nodes remembered in irg might be reachable
395 from the end node. Assure their link is set to NULL, so that
396 we can test whether new nodes have been computed. */
397 for (i = get_irg_n_anchors(irg) - 1; i >= 0; --i) {
398 ir_node *n = get_irg_anchor(irg, i);
400 set_new_node(n, NULL);
402 /* we use the block walk flag for removing Bads from Blocks ins. */
403 inc_irg_block_visited(irg);
406 copy_graph(irg, copy_node_nr);
409 old_end = get_irg_end(irg);
410 new_anchor = new_Anchor(irg);
412 for (i = get_irg_n_anchors(irg) - 1; i >= 0; --i) {
413 ir_node *n = get_irg_anchor(irg, i);
415 set_irn_n(new_anchor, i, get_new_node(n));
418 irg->anchor = new_anchor;
420 /* ensure the new anchor is placed in the endblock */
421 set_nodes_block(new_anchor, get_irg_end_block(irg));
425 * Copies all reachable nodes to a new obstack. Removes bad inputs
426 * from block nodes and the corresponding inputs from Phi nodes.
427 * Merges single exit blocks with single entry blocks and removes
429 * Adds all new nodes to a new hash table for CSE. Does not
430 * perform CSE, so the hash table might contain common subexpressions.
432 void dead_node_elimination(ir_graph *irg) {
434 #ifdef INTERPROCEDURAL_VIEW
435 int rem_ipview = get_interprocedural_view();
437 struct obstack *graveyard_obst = NULL;
438 struct obstack *rebirth_obst = NULL;
440 edges_deactivate(irg);
442 /* inform statistics that we started a dead-node elimination run */
443 hook_dead_node_elim(irg, 1);
445 /* Remember external state of current_ir_graph. */
446 rem = current_ir_graph;
447 current_ir_graph = irg;
448 #ifdef INTERPROCEDURAL_VIEW
449 set_interprocedural_view(0);
452 assert(get_irg_phase_state(irg) != phase_building);
454 /* Handle graph state */
455 free_callee_info(irg);
459 /* @@@ so far we loose loops when copying */
460 free_loop_information(irg);
462 set_irg_doms_inconsistent(irg);
464 /* A quiet place, where the old obstack can rest in peace,
465 until it will be cremated. */
466 graveyard_obst = irg->obst;
468 /* A new obstack, where the reachable nodes will be copied to. */
469 rebirth_obst = XMALLOC(struct obstack);
470 irg->obst = rebirth_obst;
471 obstack_init(irg->obst);
472 irg->last_node_idx = 0;
474 /* We also need a new value table for CSE */
475 del_identities(irg->value_table);
476 irg->value_table = new_identities();
478 /* Copy the graph from the old to the new obstack */
479 copy_graph_env(/*copy_node_nr=*/1);
481 /* Free memory from old unoptimized obstack */
482 obstack_free(graveyard_obst, 0); /* First empty the obstack ... */
483 xfree(graveyard_obst); /* ... then free it. */
485 /* inform statistics that the run is over */
486 hook_dead_node_elim(irg, 0);
488 current_ir_graph = rem;
489 #ifdef INTERPROCEDURAL_VIEW
490 set_interprocedural_view(rem_ipview);
495 * Relink bad predecessors of a block and store the old in array to the
496 * link field. This function is called by relink_bad_predecessors().
497 * The array of link field starts with the block operand at position 0.
498 * If block has bad predecessors, create a new in array without bad preds.
499 * Otherwise let in array untouched.
501 static void relink_bad_block_predecessors(ir_node *n, void *env) {
502 ir_node **new_in, *irn;
503 int i, new_irn_n, old_irn_arity, new_irn_arity = 0;
506 /* if link field of block is NULL, look for bad predecessors otherwise
507 this is already done */
508 if (is_Block(n) && get_irn_link(n) == NULL) {
509 /* save old predecessors in link field (position 0 is the block operand)*/
510 set_irn_link(n, get_irn_in(n));
512 /* count predecessors without bad nodes */
513 old_irn_arity = get_irn_arity(n);
514 for (i = 0; i < old_irn_arity; i++)
515 if (!is_Bad(get_irn_n(n, i)))
518 /* arity changing: set new predecessors without bad nodes */
519 if (new_irn_arity < old_irn_arity) {
520 /* Get new predecessor array. We do not resize the array, as we must
521 keep the old one to update Phis. */
522 new_in = NEW_ARR_D(ir_node *, current_ir_graph->obst, (new_irn_arity+1));
524 /* set new predecessors in array */
527 for (i = 0; i < old_irn_arity; i++) {
528 irn = get_irn_n(n, i);
530 new_in[new_irn_n] = irn;
531 is_backedge(n, i) ? set_backedge(n, new_irn_n-1) : set_not_backedge(n, new_irn_n-1);
535 /* ARR_SETLEN(int, n->attr.block.backedge, new_irn_arity); */
536 ARR_SHRINKLEN(n->attr.block.backedge, new_irn_arity);
538 } /* ir node has bad predecessors */
539 } /* Block is not relinked */
543 * Relinks Bad predecessors from Blocks and Phis called by walker
544 * remove_bad_predecesors(). If n is a Block, call
545 * relink_bad_block_redecessors(). If n is a Phi-node, call also the relinking
546 * function of Phi's Block. If this block has bad predecessors, relink preds
549 static void relink_bad_predecessors(ir_node *n, void *env) {
550 ir_node *block, **old_in;
551 int i, old_irn_arity, new_irn_arity;
553 /* relink bad predecessors of a block */
555 relink_bad_block_predecessors(n, env);
557 /* If Phi node relink its block and its predecessors */
559 /* Relink predecessors of phi's block */
560 block = get_nodes_block(n);
561 if (get_irn_link(block) == NULL)
562 relink_bad_block_predecessors(block, env);
564 old_in = (ir_node **)get_irn_link(block); /* Of Phi's Block */
565 old_irn_arity = ARR_LEN(old_in);
567 /* Relink Phi predecessors if count of predecessors changed */
568 if (old_irn_arity != ARR_LEN(get_irn_in(block))) {
569 /* set new predecessors in array
570 n->in[0] remains the same block */
572 for(i = 1; i < old_irn_arity; i++)
573 if (!is_Bad(old_in[i])) {
574 n->in[new_irn_arity] = n->in[i];
575 is_backedge(n, i) ? set_backedge(n, new_irn_arity) : set_not_backedge(n, new_irn_arity);
579 ARR_SETLEN(ir_node *, n->in, new_irn_arity);
580 ARR_SETLEN(int, n->attr.phi.u.backedge, new_irn_arity);
582 } /* n is a Phi node */
586 * Removes Bad Bad predecessors from Blocks and the corresponding
587 * inputs to Phi nodes as in dead_node_elimination but without
589 * On walking up set the link field to NULL, on walking down call
590 * relink_bad_predecessors() (This function stores the old in array
591 * to the link field and sets a new in array if arity of predecessors
594 void remove_bad_predecessors(ir_graph *irg) {
595 panic("Fix backedge handling first");
596 irg_walk_graph(irg, firm_clear_link, relink_bad_predecessors, NULL);
603 __)|_| | \_/ | \_/(/_ |_/\__|__
605 The following stuff implements a facility that automatically patches
606 registered ir_node pointers to the new node when a dead node elimination occurs.
609 struct _survive_dce_t {
613 hook_entry_t dead_node_elim;
614 hook_entry_t dead_node_elim_subst;
617 typedef struct _survive_dce_list_t {
618 struct _survive_dce_list_t *next;
620 } survive_dce_list_t;
622 static void dead_node_hook(void *context, ir_graph *irg, int start) {
623 survive_dce_t *sd = context;
626 /* Create a new map before the dead node elimination is performed. */
628 sd->new_places = pmap_create_ex(pmap_count(sd->places));
630 /* Patch back all nodes if dead node elimination is over and something is to be done. */
631 pmap_destroy(sd->places);
632 sd->places = sd->new_places;
633 sd->new_places = NULL;
638 * Hook called when dead node elimination replaces old by nw.
640 static void dead_node_subst_hook(void *context, ir_graph *irg, ir_node *old, ir_node *nw) {
641 survive_dce_t *sd = context;
642 survive_dce_list_t *list = pmap_get(sd->places, old);
645 /* If the node is to be patched back, write the new address to all registered locations. */
647 survive_dce_list_t *p;
649 for (p = list; p; p = p->next)
652 pmap_insert(sd->new_places, nw, list);
657 * Make a new Survive DCE environment.
659 survive_dce_t *new_survive_dce(void) {
660 survive_dce_t *res = XMALLOC(survive_dce_t);
661 obstack_init(&res->obst);
662 res->places = pmap_create();
663 res->new_places = NULL;
665 res->dead_node_elim.hook._hook_dead_node_elim = dead_node_hook;
666 res->dead_node_elim.context = res;
667 res->dead_node_elim.next = NULL;
669 res->dead_node_elim_subst.hook._hook_dead_node_elim_subst = dead_node_subst_hook;
670 res->dead_node_elim_subst.context = res;
671 res->dead_node_elim_subst.next = NULL;
673 register_hook(hook_dead_node_elim, &res->dead_node_elim);
674 register_hook(hook_dead_node_elim_subst, &res->dead_node_elim_subst);
679 * Free a Survive DCE environment.
681 void free_survive_dce(survive_dce_t *sd) {
682 obstack_free(&sd->obst, NULL);
683 pmap_destroy(sd->places);
684 unregister_hook(hook_dead_node_elim, &sd->dead_node_elim);
685 unregister_hook(hook_dead_node_elim_subst, &sd->dead_node_elim_subst);
690 * Register a node pointer to be patched upon DCE.
691 * When DCE occurs, the node pointer specified by @p place will be
692 * patched to the new address of the node it is pointing to.
694 * @param sd The Survive DCE environment.
695 * @param place The address of the node pointer.
697 void survive_dce_register_irn(survive_dce_t *sd, ir_node **place) {
698 if (*place != NULL) {
699 ir_node *irn = *place;
700 survive_dce_list_t *curr = pmap_get(sd->places, irn);
701 survive_dce_list_t *nw = obstack_alloc(&sd->obst, sizeof(nw[0]));
706 pmap_insert(sd->places, irn, nw);
710 /*--------------------------------------------------------------------*/
711 /* Functionality for inlining */
712 /*--------------------------------------------------------------------*/
715 * Copy node for inlineing. Updates attributes that change when
716 * inlineing but not for dead node elimination.
718 * Copies the node by calling copy_node() and then updates the entity if
719 * it's a local one. env must be a pointer of the frame type of the
720 * inlined procedure. The new entities must be in the link field of
723 static void copy_node_inline(ir_node *n, void *env) {
725 ir_type *frame_tp = (ir_type *)env;
729 nn = get_new_node (n);
731 if (get_entity_owner(get_Sel_entity(n)) == frame_tp) {
732 set_Sel_entity(nn, get_entity_link(get_Sel_entity(n)));
734 } else if (is_Block(n)) {
735 nn = get_new_node (n);
736 nn->attr.block.irg = current_ir_graph;
741 * Copies new predecessors of old node and move constants to
744 static void copy_preds_inline(ir_node *n, void *env) {
748 nn = skip_Id(get_new_node(n));
749 if (is_irn_constlike(nn)) {
750 /* move Constants into the start block */
751 set_nodes_block(nn, get_irg_start_block(current_ir_graph));
753 n = identify_remember(current_ir_graph->value_table, nn);
762 * Walker: checks if P_value_arg_base is used.
764 static void find_addr(ir_node *node, void *env) {
765 int *allow_inline = env;
767 ir_graph *irg = current_ir_graph;
768 if (get_Sel_ptr(node) == get_irg_frame(irg)) {
769 /* access to frame */
770 ir_entity *ent = get_Sel_entity(node);
771 if (get_entity_owner(ent) != get_irg_frame_type(irg)) {
772 /* access to value_type */
776 } else if (is_Alloc(node) && get_Alloc_where(node) == stack_alloc) {
778 * Refuse to inline alloca call unless user explicitly forced so as this
779 * may change program's memory overhead drastically when the function
780 * using alloca is called in loop. In GCC present in SPEC2000 inlining
781 * into schedule_block cause it to require 2GB of ram instead of 256MB.
783 * Sorrily this is true with our implementation also.
784 * Moreover, we cannot differentiate between alloca() and VLA yet, so this
785 * disables inlining of functions using VLA (with are completely save).
788 * - add a flag to the Alloc node for "real" alloca() calls
789 * - add a new Stack-Restore node at the end of a function using alloca()
796 * Check if we can inline a given call.
797 * Currently, we cannot inline two cases:
798 * - call with compound arguments
799 * - graphs that take the address of a parameter
801 * check these conditions here
803 static int can_inline(ir_node *call, ir_graph *called_graph) {
804 ir_type *call_type = get_Call_type(call);
805 int params, ress, i, res;
806 assert(is_Method_type(call_type));
808 params = get_method_n_params(call_type);
809 ress = get_method_n_ress(call_type);
811 /* check parameters for compound arguments */
812 for (i = 0; i < params; ++i) {
813 ir_type *p_type = get_method_param_type(call_type, i);
815 if (is_compound_type(p_type))
819 /* check results for compound arguments */
820 for (i = 0; i < ress; ++i) {
821 ir_type *r_type = get_method_res_type(call_type, i);
823 if (is_compound_type(r_type))
828 irg_walk_graph(called_graph, find_addr, NULL, &res);
834 exc_handler = 0, /**< There is a handler. */
835 exc_to_end = 1, /**< Branches to End. */
836 exc_no_handler = 2 /**< Exception handling not represented. */
839 /* Inlines a method at the given call site. */
840 int inline_method(ir_node *call, ir_graph *called_graph) {
842 ir_node *post_call, *post_bl;
843 ir_node *in[pn_Start_max];
844 ir_node *end, *end_bl, *block;
849 int arity, n_ret, n_exc, n_res, i, n, j, rem_opt, irn_arity, n_params;
850 enum exc_mode exc_handling;
851 ir_type *called_frame, *curr_frame, *mtp, *ctp;
854 irg_inline_property prop = get_irg_inline_property(called_graph);
855 unsigned long visited;
857 if (prop == irg_inline_forbidden)
860 ent = get_irg_entity(called_graph);
862 mtp = get_entity_type(ent);
863 ctp = get_Call_type(call);
864 n_params = get_method_n_params(mtp);
865 n_res = get_method_n_ress(mtp);
866 if (n_params > get_method_n_params(ctp)) {
867 /* this is a bad feature of C: without a prototype, we can
868 * call a function with less parameters than needed. Currently
869 * we don't support this, although we could use Unknown than. */
872 if (n_res != get_method_n_ress(ctp)) {
876 /* Argh, compiling C has some bad consequences:
877 * It is implementation dependent what happens in that case.
878 * We support inlining, if the bitsize of the types matches AND
879 * the same arithmetic is used. */
880 for (i = n_params - 1; i >= 0; --i) {
881 ir_type *param_tp = get_method_param_type(mtp, i);
882 ir_type *arg_tp = get_method_param_type(ctp, i);
884 if (param_tp != arg_tp) {
885 ir_mode *pmode = get_type_mode(param_tp);
886 ir_mode *amode = get_type_mode(arg_tp);
888 if (pmode == NULL || amode == NULL)
890 if (get_mode_size_bits(pmode) != get_mode_size_bits(amode))
892 if (get_mode_arithmetic(pmode) != get_mode_arithmetic(amode))
894 /* otherwise we can simply "reinterpret" the bits */
897 for (i = n_res - 1; i >= 0; --i) {
898 ir_type *decl_res_tp = get_method_res_type(mtp, i);
899 ir_type *used_res_tp = get_method_res_type(ctp, i);
901 if (decl_res_tp != used_res_tp) {
902 ir_mode *decl_mode = get_type_mode(decl_res_tp);
903 ir_mode *used_mode = get_type_mode(used_res_tp);
904 if (decl_mode == NULL || used_mode == NULL)
906 if (get_mode_size_bits(decl_mode) != get_mode_size_bits(used_mode))
908 if (get_mode_arithmetic(decl_mode) != get_mode_arithmetic(used_mode))
910 /* otherwise we can "reinterpret" the bits */
914 irg = get_irn_irg(call);
917 * We cannot inline a recursive call. The graph must be copied before
918 * the call the inline_method() using create_irg_copy().
920 if (called_graph == irg)
924 * currently, we cannot inline two cases:
925 * - call with compound arguments
926 * - graphs that take the address of a parameter
928 if (! can_inline(call, called_graph))
931 rem = current_ir_graph;
932 current_ir_graph = irg;
934 DB((dbg, LEVEL_1, "Inlining %+F(%+F) into %+F\n", call, called_graph, irg));
936 /* -- Turn off optimizations, this can cause problems when allocating new nodes. -- */
937 rem_opt = get_opt_optimize();
940 /* Handle graph state */
941 assert(get_irg_phase_state(irg) != phase_building);
942 assert(get_irg_pinned(irg) == op_pin_state_pinned);
943 assert(get_irg_pinned(called_graph) == op_pin_state_pinned);
944 set_irg_outs_inconsistent(irg);
945 set_irg_extblk_inconsistent(irg);
946 set_irg_doms_inconsistent(irg);
947 set_irg_loopinfo_inconsistent(irg);
948 set_irg_callee_info_state(irg, irg_callee_info_inconsistent);
949 set_irg_entity_usage_state(irg, ir_entity_usage_not_computed);
951 /* -- Check preconditions -- */
952 assert(is_Call(call));
954 /* here we know we WILL inline, so inform the statistics */
955 hook_inline(call, called_graph);
957 /* -- Decide how to handle exception control flow: Is there a handler
958 for the Call node, or do we branch directly to End on an exception?
960 0 There is a handler.
962 2 Exception handling not represented in Firm. -- */
964 ir_node *proj, *Mproj = NULL, *Xproj = NULL;
965 for (proj = get_irn_link(call); proj; proj = get_irn_link(proj)) {
966 long proj_nr = get_Proj_proj(proj);
967 if (proj_nr == pn_Call_X_except) Xproj = proj;
968 if (proj_nr == pn_Call_M_except) Mproj = proj;
970 if (Mproj) { assert(Xproj); exc_handling = exc_handler; } /* Mproj */
971 else if (Xproj) { exc_handling = exc_to_end; } /* !Mproj && Xproj */
972 else { exc_handling = exc_no_handler; } /* !Mproj && !Xproj */
975 /* create the argument tuple */
976 NEW_ARR_A(ir_type *, args_in, n_params);
978 block = get_nodes_block(call);
979 for (i = n_params - 1; i >= 0; --i) {
980 ir_node *arg = get_Call_param(call, i);
981 ir_type *param_tp = get_method_param_type(mtp, i);
982 ir_mode *mode = get_type_mode(param_tp);
984 if (mode != get_irn_mode(arg)) {
985 arg = new_r_Conv(irg, block, arg, mode);
991 the procedure and later replaces the Start node of the called graph.
992 Post_call is the old Call node and collects the results of the called
993 graph. Both will end up being a tuple. -- */
994 post_bl = get_nodes_block(call);
995 set_irg_current_block(irg, post_bl);
996 /* XxMxPxPxPxT of Start + parameter of Call */
997 in[pn_Start_X_initial_exec] = new_Jmp();
998 in[pn_Start_M] = get_Call_mem(call);
999 in[pn_Start_P_frame_base] = get_irg_frame(irg);
1000 in[pn_Start_P_tls] = get_irg_tls(irg);
1001 in[pn_Start_T_args] = new_Tuple(n_params, args_in);
1002 pre_call = new_Tuple(pn_Start_max, in);
1006 The new block gets the ins of the old block, pre_call and all its
1007 predecessors and all Phi nodes. -- */
1008 part_block(pre_call);
1010 /* -- Prepare state for dead node elimination -- */
1011 /* Visited flags in calling irg must be >= flag in called irg.
1012 Else walker and arity computation will not work. */
1013 if (get_irg_visited(irg) <= get_irg_visited(called_graph))
1014 set_irg_visited(irg, get_irg_visited(called_graph) + 1);
1015 if (get_irg_block_visited(irg) < get_irg_block_visited(called_graph))
1016 set_irg_block_visited(irg, get_irg_block_visited(called_graph));
1017 visited = get_irg_visited(irg);
1019 /* Set pre_call as new Start node in link field of the start node of
1020 calling graph and pre_calls block as new block for the start block
1022 Further mark these nodes so that they are not visited by the
1024 set_irn_link(get_irg_start(called_graph), pre_call);
1025 set_irn_visited(get_irg_start(called_graph), visited);
1026 set_irn_link(get_irg_start_block(called_graph), get_nodes_block(pre_call));
1027 set_irn_visited(get_irg_start_block(called_graph), visited);
1029 set_irn_link(get_irg_bad(called_graph), get_irg_bad(current_ir_graph));
1030 set_irn_visited(get_irg_bad(called_graph), visited);
1032 set_irn_link(get_irg_no_mem(called_graph), get_irg_no_mem(current_ir_graph));
1033 set_irn_visited(get_irg_no_mem(called_graph), visited);
1035 /* Initialize for compaction of in arrays */
1036 inc_irg_block_visited(irg);
1038 /* -- Replicate local entities of the called_graph -- */
1039 /* copy the entities. */
1040 irp_reserve_resources(irp, IR_RESOURCE_ENTITY_LINK);
1041 called_frame = get_irg_frame_type(called_graph);
1042 curr_frame = get_irg_frame_type(irg);
1043 for (i = 0, n = get_class_n_members(called_frame); i < n; ++i) {
1044 ir_entity *new_ent, *old_ent;
1045 old_ent = get_class_member(called_frame, i);
1046 new_ent = copy_entity_own(old_ent, curr_frame);
1047 set_entity_link(old_ent, new_ent);
1050 /* visited is > than that of called graph. With this trick visited will
1051 remain unchanged so that an outer walker, e.g., searching the call nodes
1052 to inline, calling this inline will not visit the inlined nodes. */
1053 set_irg_visited(irg, get_irg_visited(irg)-1);
1055 /* -- Performing dead node elimination inlines the graph -- */
1056 /* Copies the nodes to the obstack of current_ir_graph. Updates links to new
1058 irg_walk(get_irg_end(called_graph), copy_node_inline, copy_preds_inline,
1059 get_irg_frame_type(called_graph));
1061 irp_free_resources(irp, IR_RESOURCE_ENTITY_LINK);
1063 /* Repair called_graph */
1064 set_irg_visited(called_graph, get_irg_visited(irg));
1065 set_irg_block_visited(called_graph, get_irg_block_visited(irg));
1066 set_Block_block_visited(get_irg_start_block(called_graph), 0);
1068 /* -- Merge the end of the inlined procedure with the call site -- */
1069 /* We will turn the old Call node into a Tuple with the following
1072 0: Phi of all Memories of Return statements.
1073 1: Jmp from new Block that merges the control flow from all exception
1074 predecessors of the old end block.
1075 2: Tuple of all arguments.
1076 3: Phi of Exception memories.
1077 In case the old Call directly branches to End on an exception we don't
1078 need the block merging all exceptions nor the Phi of the exception
1082 /* -- Precompute some values -- */
1083 end_bl = get_new_node(get_irg_end_block(called_graph));
1084 end = get_new_node(get_irg_end(called_graph));
1085 arity = get_Block_n_cfgpreds(end_bl); /* arity = n_exc + n_ret */
1086 n_res = get_method_n_ress(get_Call_type(call));
1088 res_pred = XMALLOCN(ir_node*, n_res);
1089 cf_pred = XMALLOCN(ir_node*, arity);
1091 set_irg_current_block(irg, post_bl); /* just to make sure */
1093 /* -- archive keepalives -- */
1094 irn_arity = get_irn_arity(end);
1095 for (i = 0; i < irn_arity; i++) {
1096 ir_node *ka = get_End_keepalive(end, i);
1098 add_End_keepalive(get_irg_end(irg), ka);
1101 /* The new end node will die. We need not free as the in array is on the obstack:
1102 copy_node() only generated 'D' arrays. */
1104 /* -- Replace Return nodes by Jump nodes. -- */
1106 for (i = 0; i < arity; i++) {
1108 ret = get_Block_cfgpred(end_bl, i);
1109 if (is_Return(ret)) {
1110 cf_pred[n_ret] = new_r_Jmp(irg, get_nodes_block(ret));
1114 set_irn_in(post_bl, n_ret, cf_pred);
1116 /* -- Build a Tuple for all results of the method.
1117 Add Phi node if there was more than one Return. -- */
1118 turn_into_tuple(post_call, pn_Call_max);
1119 /* First the Memory-Phi */
1121 for (i = 0; i < arity; i++) {
1122 ret = get_Block_cfgpred(end_bl, i);
1123 if (is_Return(ret)) {
1124 cf_pred[n_ret] = get_Return_mem(ret);
1128 phi = new_Phi(n_ret, cf_pred, mode_M);
1129 set_Tuple_pred(call, pn_Call_M_regular, phi);
1130 /* Conserve Phi-list for further inlinings -- but might be optimized */
1131 if (get_nodes_block(phi) == post_bl) {
1132 set_irn_link(phi, get_irn_link(post_bl));
1133 set_irn_link(post_bl, phi);
1135 /* Now the real results */
1137 for (j = 0; j < n_res; j++) {
1138 ir_type *res_type = get_method_res_type(ctp, j);
1139 ir_mode *res_mode = get_type_mode(res_type);
1141 for (i = 0; i < arity; i++) {
1142 ret = get_Block_cfgpred(end_bl, i);
1143 if (is_Return(ret)) {
1144 ir_node *res = get_Return_res(ret, j);
1145 if (get_irn_mode(res) != res_mode) {
1146 ir_node *block = get_nodes_block(res);
1147 res = new_r_Conv(irg, block, res, res_mode);
1149 cf_pred[n_ret] = res;
1154 phi = new_Phi(n_ret, cf_pred, get_irn_mode(cf_pred[0]));
1158 /* Conserve Phi-list for further inlinings -- but might be optimized */
1159 if (get_nodes_block(phi) == post_bl) {
1160 set_Phi_next(phi, get_Block_phis(post_bl));
1161 set_Block_phis(post_bl, phi);
1164 set_Tuple_pred(call, pn_Call_T_result, new_Tuple(n_res, res_pred));
1166 set_Tuple_pred(call, pn_Call_T_result, new_Bad());
1168 /* handle the regular call */
1169 set_Tuple_pred(call, pn_Call_X_regular, new_Jmp());
1171 /* For now, we cannot inline calls with value_base */
1172 set_Tuple_pred(call, pn_Call_P_value_res_base, new_Bad());
1174 /* Finally the exception control flow.
1175 We have two (three) possible situations:
1176 First if the Call branches to an exception handler: We need to add a Phi node to
1177 collect the memory containing the exception objects. Further we need
1178 to add another block to get a correct representation of this Phi. To
1179 this block we add a Jmp that resolves into the X output of the Call
1180 when the Call is turned into a tuple.
1181 Second the Call branches to End, the exception is not handled. Just
1182 add all inlined exception branches to the End node.
1183 Third: there is no Exception edge at all. Handle as case two. */
1184 if (exc_handling == exc_handler) {
1186 for (i = 0; i < arity; i++) {
1188 ret = get_Block_cfgpred(end_bl, i);
1189 irn = skip_Proj(ret);
1190 if (is_fragile_op(irn) || is_Raise(irn)) {
1191 cf_pred[n_exc] = ret;
1196 ir_node *block = new_Block(n_exc, cf_pred);
1197 set_cur_block(block);
1199 set_Tuple_pred(call, pn_Call_X_except, new_Jmp());
1200 /* The Phi for the memories with the exception objects */
1202 for (i = 0; i < arity; i++) {
1204 ret = skip_Proj(get_Block_cfgpred(end_bl, i));
1206 cf_pred[n_exc] = new_r_Proj(irg, get_nodes_block(ret), ret, mode_M, 3);
1208 } else if (is_fragile_op(ret)) {
1209 /* We rely that all cfops have the memory output at the same position. */
1210 cf_pred[n_exc] = new_r_Proj(irg, get_nodes_block(ret), ret, mode_M, 0);
1212 } else if (is_Raise(ret)) {
1213 cf_pred[n_exc] = new_r_Proj(irg, get_nodes_block(ret), ret, mode_M, 1);
1217 set_Tuple_pred(call, pn_Call_M_except, new_Phi(n_exc, cf_pred, mode_M));
1219 set_Tuple_pred(call, pn_Call_X_except, new_Bad());
1220 set_Tuple_pred(call, pn_Call_M_except, new_Bad());
1223 ir_node *main_end_bl;
1224 int main_end_bl_arity;
1225 ir_node **end_preds;
1227 /* assert(exc_handling == 1 || no exceptions. ) */
1229 for (i = 0; i < arity; i++) {
1230 ir_node *ret = get_Block_cfgpred(end_bl, i);
1231 ir_node *irn = skip_Proj(ret);
1233 if (is_fragile_op(irn) || is_Raise(irn)) {
1234 cf_pred[n_exc] = ret;
1238 main_end_bl = get_irg_end_block(irg);
1239 main_end_bl_arity = get_irn_arity(main_end_bl);
1240 end_preds = XMALLOCN(ir_node*, n_exc + main_end_bl_arity);
1242 for (i = 0; i < main_end_bl_arity; ++i)
1243 end_preds[i] = get_irn_n(main_end_bl, i);
1244 for (i = 0; i < n_exc; ++i)
1245 end_preds[main_end_bl_arity + i] = cf_pred[i];
1246 set_irn_in(main_end_bl, n_exc + main_end_bl_arity, end_preds);
1247 set_Tuple_pred(call, pn_Call_X_except, new_Bad());
1248 set_Tuple_pred(call, pn_Call_M_except, new_Bad());
1254 /* -- Turn CSE back on. -- */
1255 set_optimize(rem_opt);
1256 current_ir_graph = rem;
1261 /********************************************************************/
1262 /* Apply inlining to small methods. */
1263 /********************************************************************/
1265 static struct obstack temp_obst;
1267 /** Represents a possible inlinable call in a graph. */
1268 typedef struct _call_entry {
1269 ir_node *call; /**< The Call node. */
1270 ir_graph *callee; /**< The callee IR-graph. */
1271 list_head list; /**< List head for linking the next one. */
1272 int loop_depth; /**< The loop depth of this call. */
1273 int benefice; /**< The calculated benefice of this call. */
1274 unsigned local_adr:1; /**< Set if this call gets an address of a local variable. */
1275 unsigned all_const:1; /**< Set if this call has only constant parameters. */
1279 * environment for inlining small irgs
1281 typedef struct _inline_env_t {
1282 struct obstack obst; /**< An obstack where call_entries are allocated on. */
1283 list_head calls; /**< The call entry list. */
1287 * Returns the irg called from a Call node. If the irg is not
1288 * known, NULL is returned.
1290 * @param call the call node
1292 static ir_graph *get_call_called_irg(ir_node *call) {
1295 addr = get_Call_ptr(call);
1296 if (is_Global(addr)) {
1297 ir_entity *ent = get_Global_entity(addr);
1298 return get_entity_irg(ent);
1305 * Walker: Collect all calls to known graphs inside a graph.
1307 static void collect_calls(ir_node *call, void *env) {
1309 if (is_Call(call)) {
1310 ir_graph *called_irg = get_call_called_irg(call);
1312 if (called_irg != NULL) {
1313 /* The Call node calls a locally defined method. Remember to inline. */
1314 inline_env_t *ienv = env;
1315 call_entry *entry = obstack_alloc(&ienv->obst, sizeof(*entry));
1317 entry->callee = called_irg;
1318 entry->loop_depth = 0;
1319 entry->benefice = 0;
1320 entry->local_adr = 0;
1321 entry->all_const = 0;
1323 list_add_tail(&entry->list, &ienv->calls);
1329 * Inlines all small methods at call sites where the called address comes
1330 * from a Const node that references the entity representing the called
1332 * The size argument is a rough measure for the code size of the method:
1333 * Methods where the obstack containing the firm graph is smaller than
1336 void inline_small_irgs(ir_graph *irg, int size) {
1337 ir_graph *rem = current_ir_graph;
1341 current_ir_graph = irg;
1342 /* Handle graph state */
1343 assert(get_irg_phase_state(irg) != phase_building);
1344 free_callee_info(irg);
1346 /* Find Call nodes to inline.
1347 (We can not inline during a walk of the graph, as inlining the same
1348 method several times changes the visited flag of the walked graph:
1349 after the first inlining visited of the callee equals visited of
1350 the caller. With the next inlining both are increased.) */
1351 obstack_init(&env.obst);
1352 INIT_LIST_HEAD(&env.calls);
1353 irg_walk_graph(irg, NULL, collect_calls, &env);
1355 if (! list_empty(&env.calls)) {
1356 /* There are calls to inline */
1357 ir_reserve_resources(irg, IR_RESOURCE_IRN_LINK|IR_RESOURCE_PHI_LIST);
1358 collect_phiprojs(irg);
1360 list_for_each_entry(call_entry, entry, &env.calls, list) {
1361 ir_graph *callee = entry->callee;
1362 irg_inline_property prop = get_irg_inline_property(callee);
1364 if (prop == irg_inline_forbidden || get_irg_additional_properties(callee) & mtp_property_weak) {
1365 /* do not inline forbidden / weak graphs */
1369 if (prop >= irg_inline_forced ||
1370 _obstack_memory_used(callee->obst) - (int)obstack_room(callee->obst) < size) {
1371 inline_method(entry->call, callee);
1374 ir_free_resources(irg, IR_RESOURCE_IRN_LINK|IR_RESOURCE_PHI_LIST);
1376 obstack_free(&env.obst, NULL);
1377 current_ir_graph = rem;
1381 * Environment for inlining irgs.
1384 list_head calls; /**< List of of all call nodes in this graph. */
1385 unsigned *local_weights; /**< Once allocated, the beneficial weight for transmitting local addresses. */
1386 unsigned n_nodes; /**< Number of nodes in graph except Id, Tuple, Proj, Start, End. */
1387 unsigned n_blocks; /**< Number of Blocks in graph without Start and End block. */
1388 unsigned n_nodes_orig; /**< for statistics */
1389 unsigned n_call_nodes; /**< Number of Call nodes in the graph. */
1390 unsigned n_call_nodes_orig; /**< for statistics */
1391 unsigned n_callers; /**< Number of known graphs that call this graphs. */
1392 unsigned n_callers_orig; /**< for statistics */
1393 unsigned got_inline:1; /**< Set, if at least one call inside this graph was inlined. */
1394 unsigned local_vars:1; /**< Set, if an inlined function got the address of a local variable. */
1395 unsigned recursive:1; /**< Set, if this function is self recursive. */
1399 * Allocate a new environment for inlining.
1401 static inline_irg_env *alloc_inline_irg_env(void) {
1402 inline_irg_env *env = obstack_alloc(&temp_obst, sizeof(*env));
1403 INIT_LIST_HEAD(&env->calls);
1404 env->local_weights = NULL;
1405 env->n_nodes = -2; /* do not count count Start, End */
1406 env->n_blocks = -2; /* do not count count Start, End Block */
1407 env->n_nodes_orig = -2; /* do not count Start, End */
1408 env->n_call_nodes = 0;
1409 env->n_call_nodes_orig = 0;
1411 env->n_callers_orig = 0;
1412 env->got_inline = 0;
1413 env->local_vars = 0;
1418 typedef struct walker_env {
1419 inline_irg_env *x; /**< the inline environment */
1420 char ignore_runtime; /**< the ignore runtime flag */
1421 char ignore_callers; /**< if set, do change callers data */
1425 * post-walker: collect all calls in the inline-environment
1426 * of a graph and sum some statistics.
1428 static void collect_calls2(ir_node *call, void *ctx) {
1430 inline_irg_env *x = env->x;
1431 ir_opcode code = get_irn_opcode(call);
1435 /* count meaningful nodes in irg */
1436 if (code != iro_Proj && code != iro_Tuple && code != iro_Sync) {
1437 if (code != iro_Block) {
1445 if (code != iro_Call) return;
1447 /* check, if it's a runtime call */
1448 if (env->ignore_runtime) {
1449 ir_node *symc = get_Call_ptr(call);
1451 if (is_Global(symc)) {
1452 ir_entity *ent = get_Global_entity(symc);
1454 if (get_entity_additional_properties(ent) & mtp_property_runtime)
1459 /* collect all call nodes */
1461 ++x->n_call_nodes_orig;
1463 callee = get_call_called_irg(call);
1464 if (callee != NULL) {
1465 if (! env->ignore_callers) {
1466 inline_irg_env *callee_env = get_irg_link(callee);
1467 /* count all static callers */
1468 ++callee_env->n_callers;
1469 ++callee_env->n_callers_orig;
1471 if (callee == current_ir_graph)
1474 /* link it in the list of possible inlinable entries */
1475 entry = obstack_alloc(&temp_obst, sizeof(*entry));
1477 entry->callee = callee;
1478 entry->loop_depth = get_irn_loop(get_nodes_block(call))->depth;
1479 entry->benefice = 0;
1480 entry->local_adr = 0;
1481 entry->all_const = 0;
1483 list_add_tail(&entry->list, &x->calls);
1488 * Returns TRUE if the number of callers is 0 in the irg's environment,
1489 * hence this irg is a leave.
1491 inline static int is_leave(ir_graph *irg) {
1492 inline_irg_env *env = get_irg_link(irg);
1493 return env->n_call_nodes == 0;
1497 * Returns TRUE if the number of nodes in the callee is
1498 * smaller then size in the irg's environment.
1500 inline static int is_smaller(ir_graph *callee, unsigned size) {
1501 inline_irg_env *env = get_irg_link(callee);
1502 return env->n_nodes < size;
1506 * Duplicate a call entry.
1508 * @param entry the original entry to duplicate
1509 * @param new_call the new call node
1510 * @param loop_depth_delta
1511 * delta value for the loop depth
1513 static call_entry *duplicate_call_entry(const call_entry *entry,
1514 ir_node *new_call, int loop_depth_delta) {
1515 call_entry *nentry = obstack_alloc(&temp_obst, sizeof(*nentry));
1516 nentry->call = new_call;
1517 nentry->callee = entry->callee;
1518 nentry->benefice = entry->benefice;
1519 nentry->loop_depth = entry->loop_depth + loop_depth_delta;
1520 nentry->local_adr = entry->local_adr;
1521 nentry->all_const = entry->all_const;
1527 * Append all call nodes of the source environment to the nodes of in the destination
1530 * @param dst destination environment
1531 * @param src source environment
1532 * @param loop_depth the loop depth of the call that is replaced by the src list
1534 static void append_call_list(inline_irg_env *dst, inline_irg_env *src, int loop_depth) {
1535 call_entry *entry, *nentry;
1537 /* Note that the src list points to Call nodes in the inlined graph, but
1538 we need Call nodes in our graph. Luckily the inliner leaves this information
1539 in the link field. */
1540 list_for_each_entry(call_entry, entry, &src->calls, list) {
1541 nentry = duplicate_call_entry(entry, get_irn_link(entry->call), loop_depth);
1542 list_add_tail(&nentry->list, &dst->calls);
1544 dst->n_call_nodes += src->n_call_nodes;
1545 dst->n_nodes += src->n_nodes;
1549 * Inlines small leave methods at call sites where the called address comes
1550 * from a Const node that references the entity representing the called
1552 * The size argument is a rough measure for the code size of the method:
1553 * Methods where the obstack containing the firm graph is smaller than
1556 void inline_leave_functions(unsigned maxsize, unsigned leavesize,
1557 unsigned size, int ignore_runtime)
1559 inline_irg_env *env;
1565 call_entry *entry, *next;
1566 const call_entry *centry;
1567 pmap *copied_graphs;
1568 pmap_entry *pm_entry;
1570 rem = current_ir_graph;
1571 obstack_init(&temp_obst);
1573 /* a map for the copied graphs, used to inline recursive calls */
1574 copied_graphs = pmap_create();
1576 /* extend all irgs by a temporary data structure for inlining. */
1577 n_irgs = get_irp_n_irgs();
1578 for (i = 0; i < n_irgs; ++i)
1579 set_irg_link(get_irp_irg(i), alloc_inline_irg_env());
1581 /* Pre-compute information in temporary data structure. */
1582 wenv.ignore_runtime = ignore_runtime;
1583 wenv.ignore_callers = 0;
1584 for (i = 0; i < n_irgs; ++i) {
1585 ir_graph *irg = get_irp_irg(i);
1587 assert(get_irg_phase_state(irg) != phase_building);
1588 free_callee_info(irg);
1590 assure_cf_loop(irg);
1591 wenv.x = get_irg_link(irg);
1592 irg_walk_graph(irg, NULL, collect_calls2, &wenv);
1595 /* -- and now inline. -- */
1597 /* Inline leaves recursively -- we might construct new leaves. */
1601 for (i = 0; i < n_irgs; ++i) {
1603 int phiproj_computed = 0;
1605 current_ir_graph = get_irp_irg(i);
1606 env = get_irg_link(current_ir_graph);
1608 ir_reserve_resources(current_ir_graph, IR_RESOURCE_IRN_LINK|IR_RESOURCE_PHI_LIST);
1609 list_for_each_entry_safe(call_entry, entry, next, &env->calls, list) {
1611 irg_inline_property prop;
1613 if (env->n_nodes > maxsize)
1617 callee = entry->callee;
1619 prop = get_irg_inline_property(callee);
1620 if (prop == irg_inline_forbidden || get_irg_additional_properties(callee) & mtp_property_weak) {
1621 /* do not inline forbidden / weak graphs */
1625 if (is_leave(callee) && (
1626 is_smaller(callee, leavesize) || prop >= irg_inline_forced)) {
1627 if (!phiproj_computed) {
1628 phiproj_computed = 1;
1629 collect_phiprojs(current_ir_graph);
1631 did_inline = inline_method(call, callee);
1634 inline_irg_env *callee_env = get_irg_link(callee);
1636 /* call was inlined, Phi/Projs for current graph must be recomputed */
1637 phiproj_computed = 0;
1639 /* Do some statistics */
1640 env->got_inline = 1;
1641 --env->n_call_nodes;
1642 env->n_nodes += callee_env->n_nodes;
1643 --callee_env->n_callers;
1645 /* remove this call from the list */
1646 list_del(&entry->list);
1651 ir_free_resources(current_ir_graph, IR_RESOURCE_IRN_LINK|IR_RESOURCE_PHI_LIST);
1653 } while (did_inline);
1655 /* inline other small functions. */
1656 for (i = 0; i < n_irgs; ++i) {
1658 int phiproj_computed = 0;
1660 current_ir_graph = get_irp_irg(i);
1661 env = get_irg_link(current_ir_graph);
1663 ir_reserve_resources(current_ir_graph, IR_RESOURCE_IRN_LINK|IR_RESOURCE_PHI_LIST);
1665 /* note that the list of possible calls is updated during the process */
1666 list_for_each_entry_safe(call_entry, entry, next, &env->calls, list) {
1667 irg_inline_property prop;
1672 callee = entry->callee;
1674 prop = get_irg_inline_property(callee);
1675 if (prop == irg_inline_forbidden || get_irg_additional_properties(callee) & mtp_property_weak) {
1676 /* do not inline forbidden / weak graphs */
1680 e = pmap_find(copied_graphs, callee);
1683 * Remap callee if we have a copy.
1684 * FIXME: Should we do this only for recursive Calls ?
1689 if (prop >= irg_inline_forced ||
1690 (is_smaller(callee, size) && env->n_nodes < maxsize) /* small function */) {
1691 if (current_ir_graph == callee) {
1693 * Recursive call: we cannot directly inline because we cannot walk
1694 * the graph and change it. So we have to make a copy of the graph
1698 inline_irg_env *callee_env;
1701 ir_free_resources(current_ir_graph, IR_RESOURCE_IRN_LINK|IR_RESOURCE_PHI_LIST);
1704 * No copy yet, create one.
1705 * Note that recursive methods are never leaves, so it is sufficient
1706 * to test this condition here.
1708 copy = create_irg_copy(callee);
1710 /* create_irg_copy() destroys the Proj links, recompute them */
1711 phiproj_computed = 0;
1713 ir_reserve_resources(current_ir_graph, IR_RESOURCE_IRN_LINK|IR_RESOURCE_PHI_LIST);
1715 /* allocate new environment */
1716 callee_env = alloc_inline_irg_env();
1717 set_irg_link(copy, callee_env);
1719 assure_cf_loop(copy);
1720 wenv.x = callee_env;
1721 wenv.ignore_callers = 1;
1722 irg_walk_graph(copy, NULL, collect_calls2, &wenv);
1725 * Enter the entity of the original graph. This is needed
1726 * for inline_method(). However, note that ent->irg still points
1727 * to callee, NOT to copy.
1729 set_irg_entity(copy, get_irg_entity(callee));
1731 pmap_insert(copied_graphs, callee, copy);
1734 /* we have only one caller: the original graph */
1735 callee_env->n_callers = 1;
1736 callee_env->n_callers_orig = 1;
1738 if (! phiproj_computed) {
1739 phiproj_computed = 1;
1740 collect_phiprojs(current_ir_graph);
1742 did_inline = inline_method(call, callee);
1744 inline_irg_env *callee_env = (inline_irg_env *)get_irg_link(callee);
1746 /* call was inlined, Phi/Projs for current graph must be recomputed */
1747 phiproj_computed = 0;
1749 /* callee was inline. Append it's call list. */
1750 env->got_inline = 1;
1751 --env->n_call_nodes;
1752 append_call_list(env, callee_env, entry->loop_depth);
1753 --callee_env->n_callers;
1755 /* after we have inlined callee, all called methods inside callee
1756 are now called once more */
1757 list_for_each_entry(call_entry, centry, &callee_env->calls, list) {
1758 inline_irg_env *penv = get_irg_link(centry->callee);
1762 /* remove this call from the list */
1763 list_del(&entry->list);
1768 ir_free_resources(current_ir_graph, IR_RESOURCE_IRN_LINK|IR_RESOURCE_PHI_LIST);
1771 for (i = 0; i < n_irgs; ++i) {
1772 irg = get_irp_irg(i);
1773 env = get_irg_link(irg);
1775 if (env->got_inline) {
1776 optimize_graph_df(irg);
1779 if (env->got_inline || (env->n_callers_orig != env->n_callers)) {
1780 DB((dbg, LEVEL_1, "Nodes:%3d ->%3d, calls:%3d ->%3d, callers:%3d ->%3d, -- %s\n",
1781 env->n_nodes_orig, env->n_nodes, env->n_call_nodes_orig, env->n_call_nodes,
1782 env->n_callers_orig, env->n_callers,
1783 get_entity_name(get_irg_entity(irg))));
1787 /* kill the copied graphs: we don't need them anymore */
1788 foreach_pmap(copied_graphs, pm_entry) {
1789 ir_graph *copy = pm_entry->value;
1791 /* reset the entity, otherwise it will be deleted in the next step ... */
1792 set_irg_entity(copy, NULL);
1793 free_ir_graph(copy);
1795 pmap_destroy(copied_graphs);
1797 obstack_free(&temp_obst, NULL);
1798 current_ir_graph = rem;
1802 * Calculate the parameter weights for transmitting the address of a local variable.
1804 static unsigned calc_method_local_weight(ir_node *arg) {
1806 unsigned v, weight = 0;
1808 for (i = get_irn_n_outs(arg) - 1; i >= 0; --i) {
1809 ir_node *succ = get_irn_out(arg, i);
1811 switch (get_irn_opcode(succ)) {
1814 /* Loads and Store can be removed */
1818 /* check if all args are constant */
1819 for (j = get_Sel_n_indexs(succ) - 1; j >= 0; --j) {
1820 ir_node *idx = get_Sel_index(succ, j);
1821 if (! is_Const(idx))
1824 /* Check users on this Sel. Note: if a 0 is returned here, there was
1825 some unsupported node. */
1826 v = calc_method_local_weight(succ);
1829 /* we can kill one Sel with constant indexes, this is cheap */
1833 /* when looking backward we might find Id nodes */
1834 weight += calc_method_local_weight(succ);
1837 /* unoptimized tuple */
1838 for (j = get_Tuple_n_preds(succ) - 1; j >= 0; --j) {
1839 ir_node *pred = get_Tuple_pred(succ, j);
1841 /* look for Proj(j) */
1842 for (k = get_irn_n_outs(succ) - 1; k >= 0; --k) {
1843 ir_node *succ_succ = get_irn_out(succ, k);
1844 if (is_Proj(succ_succ)) {
1845 if (get_Proj_proj(succ_succ) == j) {
1847 weight += calc_method_local_weight(succ_succ);
1850 /* this should NOT happen */
1858 /* any other node: unsupported yet or bad. */
1866 * Calculate the parameter weights for transmitting the address of a local variable.
1868 static void analyze_irg_local_weights(inline_irg_env *env, ir_graph *irg) {
1869 ir_entity *ent = get_irg_entity(irg);
1871 int nparams, i, proj_nr;
1872 ir_node *irg_args, *arg;
1874 mtp = get_entity_type(ent);
1875 nparams = get_method_n_params(mtp);
1877 /* allocate a new array. currently used as 'analysed' flag */
1878 env->local_weights = NEW_ARR_D(unsigned, &temp_obst, nparams);
1880 /* If the method haven't parameters we have nothing to do. */
1884 assure_irg_outs(irg);
1885 irg_args = get_irg_args(irg);
1886 for (i = get_irn_n_outs(irg_args) - 1; i >= 0; --i) {
1887 arg = get_irn_out(irg_args, i);
1888 proj_nr = get_Proj_proj(arg);
1889 env->local_weights[proj_nr] = calc_method_local_weight(arg);
1894 * Calculate the benefice for transmitting an local variable address.
1895 * After inlining, the local variable might be transformed into a
1896 * SSA variable by scalar_replacement().
1898 static unsigned get_method_local_adress_weight(ir_graph *callee, int pos) {
1899 inline_irg_env *env = get_irg_link(callee);
1901 if (env->local_weights != NULL) {
1902 if (pos < ARR_LEN(env->local_weights))
1903 return env->local_weights[pos];
1907 analyze_irg_local_weights(env, callee);
1909 if (pos < ARR_LEN(env->local_weights))
1910 return env->local_weights[pos];
1915 * Calculate a benefice value for inlining the given call.
1917 * @param call the call node we have to inspect
1918 * @param callee the called graph
1920 static int calc_inline_benefice(call_entry *entry, ir_graph *callee)
1922 ir_node *call = entry->call;
1923 ir_entity *ent = get_irg_entity(callee);
1927 int i, n_params, all_const;
1929 irg_inline_property prop;
1931 inline_irg_env *callee_env;
1933 prop = get_irg_inline_property(callee);
1934 if (prop == irg_inline_forbidden) {
1935 DB((dbg, LEVEL_2, "In %+F Call to %+F: inlining forbidden\n",
1937 return entry->benefice = INT_MIN;
1940 if (get_irg_additional_properties(callee) & (mtp_property_noreturn | mtp_property_weak)) {
1941 DB((dbg, LEVEL_2, "In %+F Call to %+F: not inlining noreturn or weak\n",
1943 return entry->benefice = INT_MIN;
1946 /* costs for every passed parameter */
1947 n_params = get_Call_n_params(call);
1948 mtp = get_entity_type(ent);
1949 cc = get_method_calling_convention(mtp);
1950 if (cc & cc_reg_param) {
1951 /* register parameter, smaller costs for register parameters */
1952 int max_regs = cc & ~cc_bits;
1954 if (max_regs < n_params)
1955 weight += max_regs * 2 + (n_params - max_regs) * 5;
1957 weight += n_params * 2;
1959 /* parameters are passed an stack */
1960 weight += 5 * n_params;
1963 /* constant parameters improve the benefice */
1964 frame_ptr = get_irg_frame(current_ir_graph);
1966 for (i = 0; i < n_params; ++i) {
1967 ir_node *param = get_Call_param(call, i);
1969 if (is_Const(param)) {
1970 weight += get_method_param_weight(ent, i);
1973 if (is_SymConst(param))
1974 weight += get_method_param_weight(ent, i);
1975 else if (is_Sel(param) && get_Sel_ptr(param) == frame_ptr) {
1977 * An address of a local variable is transmitted. After
1978 * inlining, scalar_replacement might be able to remove the
1979 * local variable, so honor this.
1981 v = get_method_local_adress_weight(callee, i);
1984 entry->local_adr = 1;
1988 entry->all_const = all_const;
1990 callee_env = get_irg_link(callee);
1991 if (callee_env->n_callers == 1 &&
1992 callee != current_ir_graph &&
1993 get_entity_visibility(ent) == visibility_local) {
1997 /* give a bonus for functions with one block */
1998 if (callee_env->n_blocks == 1)
1999 weight = weight * 3 / 2;
2001 /* and one for small non-recursive functions: we want them to be inlined in mostly every case */
2002 if (callee_env->n_nodes < 30 && !callee_env->recursive)
2005 /* and finally for leaves: they do not increase the register pressure
2006 because of callee safe registers */
2007 if (callee_env->n_call_nodes == 0)
2010 /** it's important to inline inner loops first */
2011 if (entry->loop_depth > 30)
2012 weight += 30 * 1024;
2014 weight += entry->loop_depth * 1024;
2017 * All arguments constant is probably a good sign, give an extra bonus
2022 return entry->benefice = weight;
2025 static ir_graph **irgs;
2026 static int last_irg;
2029 * Callgraph walker, collect all visited graphs.
2031 static void callgraph_walker(ir_graph *irg, void *data) {
2033 irgs[last_irg++] = irg;
2037 * Creates an inline order for all graphs.
2039 * @return the list of graphs.
2041 static ir_graph **create_irg_list(void) {
2042 ir_entity **free_methods;
2044 int n_irgs = get_irp_n_irgs();
2046 cgana(&arr_len, &free_methods);
2047 xfree(free_methods);
2049 compute_callgraph();
2052 irgs = XMALLOCNZ(ir_graph*, n_irgs);
2054 callgraph_walk(NULL, callgraph_walker, NULL);
2055 assert(n_irgs == last_irg);
2061 * Push a call onto the priority list if its benefice is big enough.
2063 * @param pqueue the priority queue of calls
2064 * @param call the call entry
2065 * @param inlien_threshold
2066 * the threshold value
2068 static void maybe_push_call(pqueue_t *pqueue, call_entry *call,
2069 int inline_threshold)
2071 ir_graph *callee = call->callee;
2072 irg_inline_property prop = get_irg_inline_property(callee);
2073 int benefice = calc_inline_benefice(call, callee);
2075 DB((dbg, LEVEL_2, "In %+F Call %+F to %+F has benefice %d\n",
2076 get_irn_irg(call->call), call->call, callee, benefice));
2078 if (prop < irg_inline_forced && benefice < inline_threshold) {
2082 pqueue_put(pqueue, call, benefice);
2086 * Try to inline calls into a graph.
2088 * @param irg the graph into which we inline
2089 * @param maxsize do NOT inline if the size of irg gets
2090 * bigger than this amount
2091 * @param inline_threshold
2092 * threshold value for inline decision
2093 * @param copied_graphs
2094 * map containing copied of recursive graphs
2096 static void inline_into(ir_graph *irg, unsigned maxsize,
2097 int inline_threshold, pmap *copied_graphs)
2099 int phiproj_computed = 0;
2100 inline_irg_env *env = get_irg_link(irg);
2101 call_entry *curr_call;
2105 if (env->n_call_nodes == 0)
2108 if (env->n_nodes > maxsize) {
2109 DB((dbg, LEVEL_2, "%+F: too big (%d)\n", irg, env->n_nodes));
2113 current_ir_graph = irg;
2114 ir_reserve_resources(irg, IR_RESOURCE_IRN_LINK|IR_RESOURCE_PHI_LIST);
2116 /* put irgs into the pqueue */
2117 pqueue = new_pqueue();
2119 list_for_each_entry(call_entry, curr_call, &env->calls, list) {
2120 assert(is_Call(curr_call->call));
2121 maybe_push_call(pqueue, curr_call, inline_threshold);
2124 /* note that the list of possible calls is updated during the process */
2125 while (!pqueue_empty(pqueue)) {
2127 call_entry *curr_call = pqueue_pop_front(pqueue);
2128 ir_graph *callee = curr_call->callee;
2129 ir_node *call_node = curr_call->call;
2130 inline_irg_env *callee_env = get_irg_link(callee);
2131 irg_inline_property prop = get_irg_inline_property(callee);
2133 const call_entry *centry;
2136 if ((prop < irg_inline_forced) && env->n_nodes + callee_env->n_nodes > maxsize) {
2137 DB((dbg, LEVEL_2, "%+F: too big (%d) + %+F (%d)\n", irg,
2138 env->n_nodes, callee, callee_env->n_nodes));
2142 e = pmap_find(copied_graphs, callee);
2144 int benefice = curr_call->benefice;
2146 * Reduce the weight for recursive function IFF not all arguments are const.
2147 * inlining recursive functions is rarely good.
2149 if (!curr_call->all_const)
2151 if (benefice < inline_threshold)
2155 * Remap callee if we have a copy.
2158 callee_env = get_irg_link(callee);
2161 if (current_ir_graph == callee) {
2163 * Recursive call: we cannot directly inline because we cannot
2164 * walk the graph and change it. So we have to make a copy of
2167 int benefice = curr_call->benefice;
2171 * Reduce the weight for recursive function IFF not all arguments are const.
2172 * inlining recursive functions is rarely good.
2174 if (!curr_call->all_const)
2176 if (benefice < inline_threshold)
2179 ir_free_resources(irg, IR_RESOURCE_IRN_LINK|IR_RESOURCE_PHI_LIST);
2182 * No copy yet, create one.
2183 * Note that recursive methods are never leaves, so it is
2184 * sufficient to test this condition here.
2186 copy = create_irg_copy(callee);
2188 /* create_irg_copy() destroys the Proj links, recompute them */
2189 phiproj_computed = 0;
2191 ir_reserve_resources(irg, IR_RESOURCE_IRN_LINK|IR_RESOURCE_PHI_LIST);
2193 /* allocate a new environment */
2194 callee_env = alloc_inline_irg_env();
2195 set_irg_link(copy, callee_env);
2197 assure_cf_loop(copy);
2198 wenv.x = callee_env;
2199 wenv.ignore_callers = 1;
2200 irg_walk_graph(copy, NULL, collect_calls2, &wenv);
2203 * Enter the entity of the original graph. This is needed
2204 * for inline_method(). However, note that ent->irg still points
2205 * to callee, NOT to copy.
2207 set_irg_entity(copy, get_irg_entity(callee));
2209 pmap_insert(copied_graphs, callee, copy);
2212 /* we have only one caller: the original graph */
2213 callee_env->n_callers = 1;
2214 callee_env->n_callers_orig = 1;
2216 if (! phiproj_computed) {
2217 phiproj_computed = 1;
2218 collect_phiprojs(current_ir_graph);
2220 did_inline = inline_method(call_node, callee);
2224 /* call was inlined, Phi/Projs for current graph must be recomputed */
2225 phiproj_computed = 0;
2227 /* remove it from the caller list */
2228 list_del(&curr_call->list);
2230 /* callee was inline. Append it's call list. */
2231 env->got_inline = 1;
2232 if (curr_call->local_adr)
2233 env->local_vars = 1;
2234 --env->n_call_nodes;
2236 /* we just generate a bunch of new calls */
2237 loop_depth = curr_call->loop_depth;
2238 list_for_each_entry(call_entry, centry, &callee_env->calls, list) {
2239 inline_irg_env *penv = get_irg_link(centry->callee);
2241 call_entry *new_entry;
2243 /* after we have inlined callee, all called methods inside
2244 * callee are now called once more */
2247 /* Note that the src list points to Call nodes in the inlined graph,
2248 * but we need Call nodes in our graph. Luckily the inliner leaves
2249 * this information in the link field. */
2250 new_call = get_irn_link(centry->call);
2251 assert(is_Call(new_call));
2253 new_entry = duplicate_call_entry(centry, new_call, loop_depth);
2254 list_add_tail(&new_entry->list, &env->calls);
2255 maybe_push_call(pqueue, new_entry, inline_threshold);
2258 env->n_call_nodes += callee_env->n_call_nodes;
2259 env->n_nodes += callee_env->n_nodes;
2260 --callee_env->n_callers;
2262 ir_free_resources(irg, IR_RESOURCE_IRN_LINK|IR_RESOURCE_PHI_LIST);
2267 * Heuristic inliner. Calculates a benefice value for every call and inlines
2268 * those calls with a value higher than the threshold.
2270 void inline_functions(unsigned maxsize, int inline_threshold) {
2271 inline_irg_env *env;
2275 pmap *copied_graphs;
2276 pmap_entry *pm_entry;
2279 rem = current_ir_graph;
2280 obstack_init(&temp_obst);
2282 irgs = create_irg_list();
2284 /* a map for the copied graphs, used to inline recursive calls */
2285 copied_graphs = pmap_create();
2287 /* extend all irgs by a temporary data structure for inlining. */
2288 n_irgs = get_irp_n_irgs();
2289 for (i = 0; i < n_irgs; ++i)
2290 set_irg_link(irgs[i], alloc_inline_irg_env());
2292 /* Pre-compute information in temporary data structure. */
2293 wenv.ignore_runtime = 0;
2294 wenv.ignore_callers = 0;
2295 for (i = 0; i < n_irgs; ++i) {
2296 ir_graph *irg = irgs[i];
2298 free_callee_info(irg);
2300 wenv.x = get_irg_link(irg);
2301 assure_cf_loop(irg);
2302 irg_walk_graph(irg, NULL, collect_calls2, &wenv);
2305 /* -- and now inline. -- */
2306 for (i = 0; i < n_irgs; ++i) {
2307 ir_graph *irg = irgs[i];
2309 inline_into(irg, maxsize, inline_threshold, copied_graphs);
2312 for (i = 0; i < n_irgs; ++i) {
2313 ir_graph *irg = irgs[i];
2315 env = get_irg_link(irg);
2316 if (env->got_inline) {
2317 /* this irg got calls inlined: optimize it */
2318 if (get_opt_combo()) {
2319 if (env->local_vars) {
2320 scalar_replacement_opt(irg);
2324 if (env->local_vars) {
2325 if (scalar_replacement_opt(irg)) {
2326 optimize_graph_df(irg);
2332 if (env->got_inline || (env->n_callers_orig != env->n_callers)) {
2333 DB((dbg, LEVEL_1, "Nodes:%3d ->%3d, calls:%3d ->%3d, callers:%3d ->%3d, -- %s\n",
2334 env->n_nodes_orig, env->n_nodes, env->n_call_nodes_orig, env->n_call_nodes,
2335 env->n_callers_orig, env->n_callers,
2336 get_entity_name(get_irg_entity(irg))));
2340 /* kill the copied graphs: we don't need them anymore */
2341 foreach_pmap(copied_graphs, pm_entry) {
2342 ir_graph *copy = pm_entry->value;
2344 /* reset the entity, otherwise it will be deleted in the next step ... */
2345 set_irg_entity(copy, NULL);
2346 free_ir_graph(copy);
2348 pmap_destroy(copied_graphs);
2352 obstack_free(&temp_obst, NULL);
2353 current_ir_graph = rem;
2356 void firm_init_inline(void) {
2357 FIRM_DBG_REGISTER(dbg, "firm.opt.inline");