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
34 #include "irgraph_t.h"
37 #include "iroptimize.h"
44 #include "adt/array.h"
48 #include "adt/xmalloc.h"
52 #include "irbackedge_t.h"
53 #include "opt_inline_t.h"
58 #include "analyze_irg_args.h"
59 #include "iredges_t.h"
64 DEBUG_ONLY(static firm_dbg_module_t *dbg;)
66 /*------------------------------------------------------------------*/
67 /* Routines for dead node elimination / copying garbage collection */
69 /*------------------------------------------------------------------*/
72 * Remember the new node in the old node by using a field all nodes have.
74 #define set_new_node(oldn, newn) set_irn_link(oldn, newn)
77 * Get this new node, before the old node is forgotten.
79 #define get_new_node(oldn) get_irn_link(oldn)
82 * Check if a new node was set.
84 #define has_new_node(n) (get_new_node(n) != NULL)
87 * We use the block_visited flag to mark that we have computed the
88 * number of useful predecessors for this block.
89 * Further we encode the new arity in this flag in the old blocks.
90 * Remembering the arity is useful, as it saves a lot of pointer
91 * accesses. This function is called for all Phi and Block nodes
95 compute_new_arity(ir_node *b) {
96 int i, res, irn_arity;
99 irg_v = get_irg_block_visited(current_ir_graph);
100 block_v = get_Block_block_visited(b);
101 if (block_v >= irg_v) {
102 /* we computed the number of preds for this block and saved it in the
104 return block_v - irg_v;
106 /* compute the number of good predecessors */
107 res = irn_arity = get_irn_arity(b);
108 for (i = 0; i < irn_arity; i++)
109 if (is_Bad(get_irn_n(b, i))) res--;
110 /* save it in the flag. */
111 set_Block_block_visited(b, irg_v + res);
117 * Copies the node to the new obstack. The Ins of the new node point to
118 * the predecessors on the old obstack. For block/phi nodes not all
119 * predecessors might be copied. n->link points to the new node.
120 * For Phi and Block nodes the function allocates in-arrays with an arity
121 * only for useful predecessors. The arity is determined by counting
122 * the non-bad predecessors of the block.
124 * @param n The node to be copied
125 * @param env if non-NULL, the node number attribute will be copied to the new node
127 * Note: Also used for loop unrolling.
129 static void copy_node(ir_node *n, void *env) {
132 ir_op *op = get_irn_op(n);
136 /* node copied already */
138 } else if (op == op_Block) {
140 new_arity = compute_new_arity(n);
141 n->attr.block.graph_arr = NULL;
143 block = get_nodes_block(n);
145 new_arity = compute_new_arity(block);
147 new_arity = get_irn_arity(n);
150 nn = new_ir_node(get_irn_dbg_info(n),
157 /* Copy the attributes. These might point to additional data. If this
158 was allocated on the old obstack the pointers now are dangling. This
159 frees e.g. the memory of the graph_arr allocated in new_immBlock. */
160 if (op == op_Block) {
161 /* we cannot allow blocks WITHOUT macroblock input */
162 set_Block_MacroBlock(nn, get_Block_MacroBlock(n));
164 copy_node_attr(n, nn);
168 int copy_node_nr = env != NULL;
170 /* for easier debugging, we want to copy the node numbers too */
171 nn->node_nr = n->node_nr;
177 hook_dead_node_elim_subst(current_ir_graph, n, nn);
181 * Copies new predecessors of old node to new node remembered in link.
182 * Spare the Bad predecessors of Phi and Block nodes.
184 static void copy_preds(ir_node *n, void *env) {
189 nn = get_new_node(n);
192 /* copy the macro block header */
193 ir_node *mbh = get_Block_MacroBlock(n);
196 /* this block is a macroblock header */
197 set_Block_MacroBlock(nn, nn);
199 /* get the macro block header */
200 ir_node *nmbh = get_new_node(mbh);
201 assert(nmbh != NULL);
202 set_Block_MacroBlock(nn, nmbh);
205 /* Don't copy Bad nodes. */
207 irn_arity = get_irn_arity(n);
208 for (i = 0; i < irn_arity; i++) {
209 if (! is_Bad(get_irn_n(n, i))) {
210 ir_node *pred = get_irn_n(n, i);
211 set_irn_n(nn, j, get_new_node(pred));
215 /* repair the block visited flag from above misuse. Repair it in both
216 graphs so that the old one can still be used. */
217 set_Block_block_visited(nn, 0);
218 set_Block_block_visited(n, 0);
219 /* Local optimization could not merge two subsequent blocks if
220 in array contained Bads. Now it's possible.
221 We don't call optimize_in_place as it requires
222 that the fields in ir_graph are set properly. */
223 if ((get_opt_control_flow_straightening()) &&
224 (get_Block_n_cfgpreds(nn) == 1) &&
225 is_Jmp(get_Block_cfgpred(nn, 0))) {
226 ir_node *old = get_nodes_block(get_Block_cfgpred(nn, 0));
228 /* Jmp jumps into the block it is in -- deal self cycle. */
229 assert(is_Bad(get_new_node(get_irg_bad(current_ir_graph))));
230 exchange(nn, get_new_node(get_irg_bad(current_ir_graph)));
235 } else if (is_Phi(n) && get_irn_arity(n) > 0) {
236 /* Don't copy node if corresponding predecessor in block is Bad.
237 The Block itself should not be Bad. */
238 block = get_nodes_block(n);
239 set_nodes_block(nn, get_new_node(block));
241 irn_arity = get_irn_arity(n);
242 for (i = 0; i < irn_arity; i++) {
243 if (! is_Bad(get_irn_n(block, i))) {
244 ir_node *pred = get_irn_n(n, i);
245 set_irn_n(nn, j, get_new_node(pred));
246 /*if (is_backedge(n, i)) set_backedge(nn, j);*/
250 /* If the pre walker reached this Phi after the post walker visited the
251 block block_visited is > 0. */
252 set_Block_block_visited(get_nodes_block(n), 0);
253 /* Compacting the Phi's ins might generate Phis with only one
255 if (get_irn_arity(nn) == 1)
256 exchange(nn, get_irn_n(nn, 0));
258 irn_arity = get_irn_arity(n);
259 for (i = -1; i < irn_arity; i++)
260 set_irn_n(nn, i, get_new_node(get_irn_n(n, i)));
262 /* Now the new node is complete. We can add it to the hash table for CSE.
263 @@@ inlining aborts if we identify End. Why? */
265 add_identities(current_ir_graph->value_table, nn);
269 * Copies the graph recursively, compacts the keep-alives of the end node.
271 * @param irg the graph to be copied
272 * @param copy_node_nr If non-zero, the node number will be copied
274 static void copy_graph(ir_graph *irg, int copy_node_nr) {
275 ir_node *oe, *ne, *ob, *nb, *om, *nm; /* old end, new end, old bad, new bad, old NoMem, new NoMem */
276 ir_node *ka; /* keep alive */
280 /* Some nodes must be copied by hand, sigh */
281 vfl = get_irg_visited(irg);
282 set_irg_visited(irg, vfl + 1);
284 oe = get_irg_end(irg);
285 mark_irn_visited(oe);
286 /* copy the end node by hand, allocate dynamic in array! */
287 ne = new_ir_node(get_irn_dbg_info(oe),
294 /* Copy the attributes. Well, there might be some in the future... */
295 copy_node_attr(oe, ne);
296 set_new_node(oe, ne);
298 /* copy the Bad node */
299 ob = get_irg_bad(irg);
300 mark_irn_visited(ob);
301 nb = new_ir_node(get_irn_dbg_info(ob),
308 copy_node_attr(ob, nb);
309 set_new_node(ob, nb);
311 /* copy the NoMem node */
312 om = get_irg_no_mem(irg);
313 mark_irn_visited(om);
314 nm = new_ir_node(get_irn_dbg_info(om),
321 copy_node_attr(om, nm);
322 set_new_node(om, nm);
324 /* copy the live nodes */
325 set_irg_visited(irg, vfl);
326 irg_walk(get_nodes_block(oe), copy_node, copy_preds, INT_TO_PTR(copy_node_nr));
328 /* Note: from yet, the visited flag of the graph is equal to vfl + 1 */
330 /* visit the anchors as well */
331 for (i = get_irg_n_anchors(irg) - 1; i >= 0; --i) {
332 ir_node *n = get_irg_anchor(irg, i);
334 if (n && (get_irn_visited(n) <= vfl)) {
335 set_irg_visited(irg, vfl);
336 irg_walk(n, copy_node, copy_preds, INT_TO_PTR(copy_node_nr));
340 /* copy_preds for the end node ... */
341 set_nodes_block(ne, get_new_node(get_nodes_block(oe)));
343 /*- ... and now the keep alives. -*/
344 /* First pick the not marked block nodes and walk them. We must pick these
345 first as else we will oversee blocks reachable from Phis. */
346 irn_arity = get_End_n_keepalives(oe);
347 for (i = 0; i < irn_arity; i++) {
348 ka = get_End_keepalive(oe, i);
350 if (get_irn_visited(ka) <= vfl) {
351 /* We must keep the block alive and copy everything reachable */
352 set_irg_visited(irg, vfl);
353 irg_walk(ka, copy_node, copy_preds, INT_TO_PTR(copy_node_nr));
355 add_End_keepalive(ne, get_new_node(ka));
359 /* Now pick other nodes. Here we will keep all! */
360 irn_arity = get_End_n_keepalives(oe);
361 for (i = 0; i < irn_arity; i++) {
362 ka = get_End_keepalive(oe, i);
364 if (get_irn_visited(ka) <= vfl) {
365 /* We didn't copy the node yet. */
366 set_irg_visited(irg, vfl);
367 irg_walk(ka, copy_node, copy_preds, INT_TO_PTR(copy_node_nr));
369 add_End_keepalive(ne, get_new_node(ka));
373 /* start block sometimes only reached after keep alives */
374 set_nodes_block(nb, get_new_node(get_nodes_block(ob)));
375 set_nodes_block(nm, get_new_node(get_nodes_block(om)));
379 * Copies the graph reachable from current_ir_graph->end to the obstack
380 * in current_ir_graph and fixes the environment.
381 * Then fixes the fields in current_ir_graph containing nodes of the
384 * @param copy_node_nr If non-zero, the node number will be copied
387 copy_graph_env(int copy_node_nr) {
388 ir_graph *irg = current_ir_graph;
389 ir_node *old_end, *new_anchor;
392 /* remove end_except and end_reg nodes */
393 old_end = get_irg_end(irg);
394 set_irg_end_except (irg, old_end);
395 set_irg_end_reg (irg, old_end);
397 /* Not all nodes remembered in irg might be reachable
398 from the end node. Assure their link is set to NULL, so that
399 we can test whether new nodes have been computed. */
400 for (i = get_irg_n_anchors(irg) - 1; i >= 0; --i) {
401 ir_node *n = get_irg_anchor(irg, i);
403 set_new_node(n, NULL);
405 /* we use the block walk flag for removing Bads from Blocks ins. */
406 inc_irg_block_visited(irg);
409 copy_graph(irg, copy_node_nr);
412 old_end = get_irg_end(irg);
413 new_anchor = new_Anchor(irg);
415 for (i = get_irg_n_anchors(irg) - 1; i >= 0; --i) {
416 ir_node *n = get_irg_anchor(irg, i);
418 set_irn_n(new_anchor, i, get_new_node(n));
421 irg->anchor = new_anchor;
423 /* ensure the new anchor is placed in the endblock */
424 set_nodes_block(new_anchor, get_irg_end_block(irg));
428 * Copies all reachable nodes to a new obstack. Removes bad inputs
429 * from block nodes and the corresponding inputs from Phi nodes.
430 * Merges single exit blocks with single entry blocks and removes
432 * Adds all new nodes to a new hash table for CSE. Does not
433 * perform CSE, so the hash table might contain common subexpressions.
435 void dead_node_elimination(ir_graph *irg) {
437 #ifdef INTERPROCEDURAL_VIEW
438 int rem_ipview = get_interprocedural_view();
440 struct obstack *graveyard_obst = NULL;
441 struct obstack *rebirth_obst = NULL;
442 assert(! edges_activated(irg) && "dead node elimination requires disabled edges");
444 /* inform statistics that we started a dead-node elimination run */
445 hook_dead_node_elim(irg, 1);
447 /* Remember external state of current_ir_graph. */
448 rem = current_ir_graph;
449 current_ir_graph = irg;
450 #ifdef INTERPROCEDURAL_VIEW
451 set_interprocedural_view(0);
454 assert(get_irg_phase_state(irg) != phase_building);
456 /* Handle graph state */
457 free_callee_info(irg);
461 /* @@@ so far we loose loops when copying */
462 free_loop_information(irg);
464 set_irg_doms_inconsistent(irg);
466 /* A quiet place, where the old obstack can rest in peace,
467 until it will be cremated. */
468 graveyard_obst = irg->obst;
470 /* A new obstack, where the reachable nodes will be copied to. */
471 rebirth_obst = xmalloc(sizeof(*rebirth_obst));
472 irg->obst = rebirth_obst;
473 obstack_init(irg->obst);
474 irg->last_node_idx = 0;
476 /* We also need a new value table for CSE */
477 del_identities(irg->value_table);
478 irg->value_table = new_identities();
480 /* Copy the graph from the old to the new obstack */
481 copy_graph_env(/*copy_node_nr=*/1);
483 /* Free memory from old unoptimized obstack */
484 obstack_free(graveyard_obst, 0); /* First empty the obstack ... */
485 xfree(graveyard_obst); /* ... then free it. */
487 /* inform statistics that the run is over */
488 hook_dead_node_elim(irg, 0);
490 current_ir_graph = rem;
491 #ifdef INTERPROCEDURAL_VIEW
492 set_interprocedural_view(rem_ipview);
497 * Relink bad predecessors of a block and store the old in array to the
498 * link field. This function is called by relink_bad_predecessors().
499 * The array of link field starts with the block operand at position 0.
500 * If block has bad predecessors, create a new in array without bad preds.
501 * Otherwise let in array untouched.
503 static void relink_bad_block_predecessors(ir_node *n, void *env) {
504 ir_node **new_in, *irn;
505 int i, new_irn_n, old_irn_arity, new_irn_arity = 0;
508 /* if link field of block is NULL, look for bad predecessors otherwise
509 this is already done */
510 if (is_Block(n) && get_irn_link(n) == NULL) {
511 /* save old predecessors in link field (position 0 is the block operand)*/
512 set_irn_link(n, get_irn_in(n));
514 /* count predecessors without bad nodes */
515 old_irn_arity = get_irn_arity(n);
516 for (i = 0; i < old_irn_arity; i++)
517 if (!is_Bad(get_irn_n(n, i)))
520 /* arity changing: set new predecessors without bad nodes */
521 if (new_irn_arity < old_irn_arity) {
522 /* Get new predecessor array. We do not resize the array, as we must
523 keep the old one to update Phis. */
524 new_in = NEW_ARR_D(ir_node *, current_ir_graph->obst, (new_irn_arity+1));
526 /* set new predecessors in array */
529 for (i = 0; i < old_irn_arity; i++) {
530 irn = get_irn_n(n, i);
532 new_in[new_irn_n] = irn;
533 is_backedge(n, i) ? set_backedge(n, new_irn_n-1) : set_not_backedge(n, new_irn_n-1);
537 /* ARR_SETLEN(int, n->attr.block.backedge, new_irn_arity); */
538 ARR_SHRINKLEN(n->attr.block.backedge, new_irn_arity);
540 } /* ir node has bad predecessors */
541 } /* Block is not relinked */
545 * Relinks Bad predecessors from Blocks and Phis called by walker
546 * remove_bad_predecesors(). If n is a Block, call
547 * relink_bad_block_redecessors(). If n is a Phi-node, call also the relinking
548 * function of Phi's Block. If this block has bad predecessors, relink preds
551 static void relink_bad_predecessors(ir_node *n, void *env) {
552 ir_node *block, **old_in;
553 int i, old_irn_arity, new_irn_arity;
555 /* relink bad predecessors of a block */
557 relink_bad_block_predecessors(n, env);
559 /* If Phi node relink its block and its predecessors */
561 /* Relink predecessors of phi's block */
562 block = get_nodes_block(n);
563 if (get_irn_link(block) == NULL)
564 relink_bad_block_predecessors(block, env);
566 old_in = (ir_node **)get_irn_link(block); /* Of Phi's Block */
567 old_irn_arity = ARR_LEN(old_in);
569 /* Relink Phi predecessors if count of predecessors changed */
570 if (old_irn_arity != ARR_LEN(get_irn_in(block))) {
571 /* set new predecessors in array
572 n->in[0] remains the same block */
574 for(i = 1; i < old_irn_arity; i++)
575 if (!is_Bad(old_in[i])) {
576 n->in[new_irn_arity] = n->in[i];
577 is_backedge(n, i) ? set_backedge(n, new_irn_arity) : set_not_backedge(n, new_irn_arity);
581 ARR_SETLEN(ir_node *, n->in, new_irn_arity);
582 ARR_SETLEN(int, n->attr.phi.u.backedge, new_irn_arity);
584 } /* n is a Phi node */
588 * Removes Bad Bad predecessors from Blocks and the corresponding
589 * inputs to Phi nodes as in dead_node_elimination but without
591 * On walking up set the link field to NULL, on walking down call
592 * relink_bad_predecessors() (This function stores the old in array
593 * to the link field and sets a new in array if arity of predecessors
596 void remove_bad_predecessors(ir_graph *irg) {
597 panic("Fix backedge handling first");
598 irg_walk_graph(irg, firm_clear_link, relink_bad_predecessors, NULL);
605 __)|_| | \_/ | \_/(/_ |_/\__|__
607 The following stuff implements a facility that automatically patches
608 registered ir_node pointers to the new node when a dead node elimination occurs.
611 struct _survive_dce_t {
615 hook_entry_t dead_node_elim;
616 hook_entry_t dead_node_elim_subst;
619 typedef struct _survive_dce_list_t {
620 struct _survive_dce_list_t *next;
622 } survive_dce_list_t;
624 static void dead_node_hook(void *context, ir_graph *irg, int start) {
625 survive_dce_t *sd = context;
628 /* Create a new map before the dead node elimination is performed. */
630 sd->new_places = pmap_create_ex(pmap_count(sd->places));
632 /* Patch back all nodes if dead node elimination is over and something is to be done. */
633 pmap_destroy(sd->places);
634 sd->places = sd->new_places;
635 sd->new_places = NULL;
640 * Hook called when dead node elimination replaces old by nw.
642 static void dead_node_subst_hook(void *context, ir_graph *irg, ir_node *old, ir_node *nw) {
643 survive_dce_t *sd = context;
644 survive_dce_list_t *list = pmap_get(sd->places, old);
647 /* If the node is to be patched back, write the new address to all registered locations. */
649 survive_dce_list_t *p;
651 for (p = list; p; p = p->next)
654 pmap_insert(sd->new_places, nw, list);
659 * Make a new Survive DCE environment.
661 survive_dce_t *new_survive_dce(void) {
662 survive_dce_t *res = xmalloc(sizeof(res[0]));
663 obstack_init(&res->obst);
664 res->places = pmap_create();
665 res->new_places = NULL;
667 res->dead_node_elim.hook._hook_dead_node_elim = dead_node_hook;
668 res->dead_node_elim.context = res;
669 res->dead_node_elim.next = NULL;
671 res->dead_node_elim_subst.hook._hook_dead_node_elim_subst = dead_node_subst_hook;
672 res->dead_node_elim_subst.context = res;
673 res->dead_node_elim_subst.next = NULL;
675 #ifndef FIRM_ENABLE_HOOKS
676 assert(0 && "need hooks enabled");
679 register_hook(hook_dead_node_elim, &res->dead_node_elim);
680 register_hook(hook_dead_node_elim_subst, &res->dead_node_elim_subst);
685 * Free a Survive DCE environment.
687 void free_survive_dce(survive_dce_t *sd) {
688 obstack_free(&sd->obst, NULL);
689 pmap_destroy(sd->places);
690 unregister_hook(hook_dead_node_elim, &sd->dead_node_elim);
691 unregister_hook(hook_dead_node_elim_subst, &sd->dead_node_elim_subst);
696 * Register a node pointer to be patched upon DCE.
697 * When DCE occurs, the node pointer specified by @p place will be
698 * patched to the new address of the node it is pointing to.
700 * @param sd The Survive DCE environment.
701 * @param place The address of the node pointer.
703 void survive_dce_register_irn(survive_dce_t *sd, ir_node **place) {
704 if (*place != NULL) {
705 ir_node *irn = *place;
706 survive_dce_list_t *curr = pmap_get(sd->places, irn);
707 survive_dce_list_t *nw = obstack_alloc(&sd->obst, sizeof(nw[0]));
712 pmap_insert(sd->places, irn, nw);
716 /*--------------------------------------------------------------------*/
717 /* Functionality for inlining */
718 /*--------------------------------------------------------------------*/
721 * Copy node for inlineing. Updates attributes that change when
722 * inlineing but not for dead node elimination.
724 * Copies the node by calling copy_node() and then updates the entity if
725 * it's a local one. env must be a pointer of the frame type of the
726 * inlined procedure. The new entities must be in the link field of
730 copy_node_inline(ir_node *n, void *env) {
732 ir_type *frame_tp = (ir_type *)env;
736 nn = get_new_node (n);
738 if (get_entity_owner(get_Sel_entity(n)) == frame_tp) {
739 set_Sel_entity(nn, get_entity_link(get_Sel_entity(n)));
741 } else if (is_Block(n)) {
742 nn = get_new_node (n);
743 nn->attr.block.irg = current_ir_graph;
748 * Walker: checks if P_value_arg_base is used.
750 static void find_addr(ir_node *node, void *env) {
751 int *allow_inline = env;
753 is_Start(get_Proj_pred(node)) &&
754 get_Proj_proj(node) == pn_Start_P_value_arg_base) {
756 } else if (is_Alloc(node) && get_Alloc_where(node) == stack_alloc) {
758 * Refuse to inline alloca call unless user explicitly forced so as this
759 * may change program's memory overhead drastically when the function
760 * using alloca is called in loop. In GCC present in SPEC2000 inlining
761 * into schedule_block cause it to require 2GB of ram instead of 256MB.
763 * Sorryly this is true with our implementation also.
764 * Moreover, we cannot differentiate between alloca() and VLA yet, so this
765 * disables inlining of functions using VLA (with are completely save).
768 * - add a flag to the Alloc node for "real" alloca() calls
769 * - add a new Stack-Restore node at the end of a function using alloca()
776 * Check if we can inline a given call.
777 * Currently, we cannot inline two cases:
778 * - call with compound arguments
779 * - graphs that take the address of a parameter
781 * check these conditions here
783 static int can_inline(ir_node *call, ir_graph *called_graph) {
784 ir_type *call_type = get_Call_type(call);
785 int params, ress, i, res;
786 assert(is_Method_type(call_type));
788 params = get_method_n_params(call_type);
789 ress = get_method_n_ress(call_type);
791 /* check parameters for compound arguments */
792 for (i = 0; i < params; ++i) {
793 ir_type *p_type = get_method_param_type(call_type, i);
795 if (is_compound_type(p_type))
799 /* check results for compound arguments */
800 for (i = 0; i < ress; ++i) {
801 ir_type *r_type = get_method_res_type(call_type, i);
803 if (is_compound_type(r_type))
808 irg_walk_graph(called_graph, find_addr, NULL, &res);
814 exc_handler = 0, /**< There is a handler. */
815 exc_to_end = 1, /**< Branches to End. */
816 exc_no_handler = 2 /**< Exception handling not represented. */
819 /* Inlines a method at the given call site. */
820 int inline_method(ir_node *call, ir_graph *called_graph) {
822 ir_node *post_call, *post_bl;
823 ir_node *in[pn_Start_max];
824 ir_node *end, *end_bl;
828 int arity, n_ret, n_exc, n_res, i, n, j, rem_opt, irn_arity;
829 enum exc_mode exc_handling;
830 ir_type *called_frame, *curr_frame;
833 irg_inline_property prop = get_irg_inline_property(called_graph);
835 if (prop == irg_inline_forbidden)
838 ent = get_irg_entity(called_graph);
840 /* Do not inline variadic functions. */
841 if (get_method_variadicity(get_entity_type(ent)) == variadicity_variadic)
844 assert(get_method_n_params(get_entity_type(ent)) ==
845 get_method_n_params(get_Call_type(call)));
847 irg = get_irn_irg(call);
850 * We cannot inline a recursive call. The graph must be copied before
851 * the call the inline_method() using create_irg_copy().
853 if (called_graph == irg)
857 * currently, we cannot inline two cases:
858 * - call with compound arguments
859 * - graphs that take the address of a parameter
861 if (! can_inline(call, called_graph))
864 rem = current_ir_graph;
865 current_ir_graph = irg;
867 DB((dbg, SET_LEVEL_1, "Inlining %+F(%+F) into %+F\n", call, called_graph, irg));
869 /* -- Turn off optimizations, this can cause problems when allocating new nodes. -- */
870 rem_opt = get_opt_optimize();
873 /* Handle graph state */
874 assert(get_irg_phase_state(irg) != phase_building);
875 assert(get_irg_pinned(irg) == op_pin_state_pinned);
876 assert(get_irg_pinned(called_graph) == op_pin_state_pinned);
877 set_irg_outs_inconsistent(irg);
878 set_irg_extblk_inconsistent(irg);
879 set_irg_doms_inconsistent(irg);
880 set_irg_loopinfo_inconsistent(irg);
881 set_irg_callee_info_state(irg, irg_callee_info_inconsistent);
883 /* -- Check preconditions -- */
884 assert(is_Call(call));
886 /* here we know we WILL inline, so inform the statistics */
887 hook_inline(call, called_graph);
889 /* -- Decide how to handle exception control flow: Is there a handler
890 for the Call node, or do we branch directly to End on an exception?
892 0 There is a handler.
894 2 Exception handling not represented in Firm. -- */
896 ir_node *proj, *Mproj = NULL, *Xproj = NULL;
897 for (proj = get_irn_link(call); proj; proj = get_irn_link(proj)) {
898 long proj_nr = get_Proj_proj(proj);
899 if (proj_nr == pn_Call_X_except) Xproj = proj;
900 if (proj_nr == pn_Call_M_except) Mproj = proj;
902 if (Mproj) { assert(Xproj); exc_handling = exc_handler; } /* Mproj */
903 else if (Xproj) { exc_handling = exc_to_end; } /* !Mproj && Xproj */
904 else { exc_handling = exc_no_handler; } /* !Mproj && !Xproj */
908 the procedure and later replaces the Start node of the called graph.
909 Post_call is the old Call node and collects the results of the called
910 graph. Both will end up being a tuple. -- */
911 post_bl = get_nodes_block(call);
912 set_irg_current_block(irg, post_bl);
913 /* XxMxPxPxPxT of Start + parameter of Call */
914 in[pn_Start_X_initial_exec] = new_Jmp();
915 in[pn_Start_M] = get_Call_mem(call);
916 in[pn_Start_P_frame_base] = get_irg_frame(irg);
917 in[pn_Start_P_globals] = get_irg_globals(irg);
918 in[pn_Start_P_tls] = get_irg_tls(irg);
919 in[pn_Start_T_args] = new_Tuple(get_Call_n_params(call), get_Call_param_arr(call));
920 /* in[pn_Start_P_value_arg_base] = ??? */
921 assert(pn_Start_P_value_arg_base == pn_Start_max - 1 && "pn_Start_P_value_arg_base not supported, fix");
922 pre_call = new_Tuple(pn_Start_max - 1, in);
926 The new block gets the ins of the old block, pre_call and all its
927 predecessors and all Phi nodes. -- */
928 part_block(pre_call);
930 /* -- Prepare state for dead node elimination -- */
931 /* Visited flags in calling irg must be >= flag in called irg.
932 Else walker and arity computation will not work. */
933 if (get_irg_visited(irg) <= get_irg_visited(called_graph))
934 set_irg_visited(irg, get_irg_visited(called_graph)+1);
935 if (get_irg_block_visited(irg) < get_irg_block_visited(called_graph))
936 set_irg_block_visited(irg, get_irg_block_visited(called_graph));
937 /* Set pre_call as new Start node in link field of the start node of
938 calling graph and pre_calls block as new block for the start block
940 Further mark these nodes so that they are not visited by the
942 set_irn_link(get_irg_start(called_graph), pre_call);
943 set_irn_visited(get_irg_start(called_graph), get_irg_visited(irg));
944 set_irn_link(get_irg_start_block(called_graph), get_nodes_block(pre_call));
945 set_irn_visited(get_irg_start_block(called_graph), get_irg_visited(irg));
946 set_irn_link(get_irg_bad(called_graph), get_irg_bad(irg));
947 set_irn_visited(get_irg_bad(called_graph), get_irg_visited(irg));
949 /* Initialize for compaction of in arrays */
950 inc_irg_block_visited(irg);
952 /* -- Replicate local entities of the called_graph -- */
953 /* copy the entities. */
954 called_frame = get_irg_frame_type(called_graph);
955 curr_frame = get_irg_frame_type(irg);
956 for (i = 0, n = get_class_n_members(called_frame); i < n; ++i) {
957 ir_entity *new_ent, *old_ent;
958 old_ent = get_class_member(called_frame, i);
959 new_ent = copy_entity_own(old_ent, curr_frame);
960 set_entity_link(old_ent, new_ent);
963 /* visited is > than that of called graph. With this trick visited will
964 remain unchanged so that an outer walker, e.g., searching the call nodes
965 to inline, calling this inline will not visit the inlined nodes. */
966 set_irg_visited(irg, get_irg_visited(irg)-1);
968 /* -- Performing dead node elimination inlines the graph -- */
969 /* Copies the nodes to the obstack of current_ir_graph. Updates links to new
971 irg_walk(get_irg_end(called_graph), copy_node_inline, copy_preds,
972 get_irg_frame_type(called_graph));
974 /* Repair called_graph */
975 set_irg_visited(called_graph, get_irg_visited(irg));
976 set_irg_block_visited(called_graph, get_irg_block_visited(irg));
977 set_Block_block_visited(get_irg_start_block(called_graph), 0);
979 /* -- Merge the end of the inlined procedure with the call site -- */
980 /* We will turn the old Call node into a Tuple with the following
983 0: Phi of all Memories of Return statements.
984 1: Jmp from new Block that merges the control flow from all exception
985 predecessors of the old end block.
986 2: Tuple of all arguments.
987 3: Phi of Exception memories.
988 In case the old Call directly branches to End on an exception we don't
989 need the block merging all exceptions nor the Phi of the exception
993 /* -- Precompute some values -- */
994 end_bl = get_new_node(get_irg_end_block(called_graph));
995 end = get_new_node(get_irg_end(called_graph));
996 arity = get_irn_arity(end_bl); /* arity = n_exc + n_ret */
997 n_res = get_method_n_ress(get_Call_type(call));
999 res_pred = xmalloc(n_res * sizeof(*res_pred));
1000 cf_pred = xmalloc(arity * sizeof(*res_pred));
1002 set_irg_current_block(irg, post_bl); /* just to make sure */
1004 /* -- archive keepalives -- */
1005 irn_arity = get_irn_arity(end);
1006 for (i = 0; i < irn_arity; i++) {
1007 ir_node *ka = get_End_keepalive(end, i);
1009 add_End_keepalive(get_irg_end(irg), ka);
1012 /* The new end node will die. We need not free as the in array is on the obstack:
1013 copy_node() only generated 'D' arrays. */
1015 /* -- Replace Return nodes by Jump nodes. -- */
1017 for (i = 0; i < arity; i++) {
1019 ret = get_irn_n(end_bl, i);
1020 if (is_Return(ret)) {
1021 cf_pred[n_ret] = new_r_Jmp(irg, get_nodes_block(ret));
1025 set_irn_in(post_bl, n_ret, cf_pred);
1027 /* -- Build a Tuple for all results of the method.
1028 Add Phi node if there was more than one Return. -- */
1029 turn_into_tuple(post_call, pn_Call_max);
1030 /* First the Memory-Phi */
1032 for (i = 0; i < arity; i++) {
1033 ret = get_irn_n(end_bl, i);
1034 if (is_Return(ret)) {
1035 cf_pred[n_ret] = get_Return_mem(ret);
1039 phi = new_Phi(n_ret, cf_pred, mode_M);
1040 set_Tuple_pred(call, pn_Call_M_regular, phi);
1041 /* Conserve Phi-list for further inlinings -- but might be optimized */
1042 if (get_nodes_block(phi) == post_bl) {
1043 set_irn_link(phi, get_irn_link(post_bl));
1044 set_irn_link(post_bl, phi);
1046 /* Now the real results */
1048 for (j = 0; j < n_res; j++) {
1050 for (i = 0; i < arity; i++) {
1051 ret = get_irn_n(end_bl, i);
1052 if (is_Return(ret)) {
1053 cf_pred[n_ret] = get_Return_res(ret, j);
1058 phi = new_Phi(n_ret, cf_pred, get_irn_mode(cf_pred[0]));
1062 /* Conserve Phi-list for further inlinings -- but might be optimized */
1063 if (get_nodes_block(phi) == post_bl) {
1064 set_Phi_next(phi, get_Block_phis(post_bl));
1065 set_Block_phis(post_bl, phi);
1068 set_Tuple_pred(call, pn_Call_T_result, new_Tuple(n_res, res_pred));
1070 set_Tuple_pred(call, pn_Call_T_result, new_Bad());
1072 /* handle the regular call */
1073 set_Tuple_pred(call, pn_Call_X_regular, new_Jmp());
1075 /* For now, we cannot inline calls with value_base */
1076 set_Tuple_pred(call, pn_Call_P_value_res_base, new_Bad());
1078 /* Finally the exception control flow.
1079 We have two (three) possible situations:
1080 First if the Call branches to an exception handler: We need to add a Phi node to
1081 collect the memory containing the exception objects. Further we need
1082 to add another block to get a correct representation of this Phi. To
1083 this block we add a Jmp that resolves into the X output of the Call
1084 when the Call is turned into a tuple.
1085 Second the Call branches to End, the exception is not handled. Just
1086 add all inlined exception branches to the End node.
1087 Third: there is no Exception edge at all. Handle as case two. */
1088 if (exc_handling == exc_handler) {
1090 for (i = 0; i < arity; i++) {
1092 ret = get_irn_n(end_bl, i);
1093 irn = skip_Proj(ret);
1094 if (is_fragile_op(irn) || is_Raise(irn)) {
1095 cf_pred[n_exc] = ret;
1100 new_Block(n_exc, cf_pred); /* watch it: current_block is changed! */
1101 set_Tuple_pred(call, pn_Call_X_except, new_Jmp());
1102 /* The Phi for the memories with the exception objects */
1104 for (i = 0; i < arity; i++) {
1106 ret = skip_Proj(get_irn_n(end_bl, i));
1108 cf_pred[n_exc] = new_r_Proj(irg, get_nodes_block(ret), ret, mode_M, 3);
1110 } else if (is_fragile_op(ret)) {
1111 /* We rely that all cfops have the memory output at the same position. */
1112 cf_pred[n_exc] = new_r_Proj(irg, get_nodes_block(ret), ret, mode_M, 0);
1114 } else if (is_Raise(ret)) {
1115 cf_pred[n_exc] = new_r_Proj(irg, get_nodes_block(ret), ret, mode_M, 1);
1119 set_Tuple_pred(call, pn_Call_M_except, new_Phi(n_exc, cf_pred, mode_M));
1121 set_Tuple_pred(call, pn_Call_X_except, new_Bad());
1122 set_Tuple_pred(call, pn_Call_M_except, new_Bad());
1125 ir_node *main_end_bl;
1126 int main_end_bl_arity;
1127 ir_node **end_preds;
1129 /* assert(exc_handling == 1 || no exceptions. ) */
1131 for (i = 0; i < arity; i++) {
1132 ir_node *ret = get_irn_n(end_bl, i);
1133 ir_node *irn = skip_Proj(ret);
1135 if (is_fragile_op(irn) || is_Raise(irn)) {
1136 cf_pred[n_exc] = ret;
1140 main_end_bl = get_irg_end_block(irg);
1141 main_end_bl_arity = get_irn_arity(main_end_bl);
1142 end_preds = xmalloc((n_exc + main_end_bl_arity) * sizeof(*end_preds));
1144 for (i = 0; i < main_end_bl_arity; ++i)
1145 end_preds[i] = get_irn_n(main_end_bl, i);
1146 for (i = 0; i < n_exc; ++i)
1147 end_preds[main_end_bl_arity + i] = cf_pred[i];
1148 set_irn_in(main_end_bl, n_exc + main_end_bl_arity, end_preds);
1149 set_Tuple_pred(call, pn_Call_X_except, new_Bad());
1150 set_Tuple_pred(call, pn_Call_M_except, new_Bad());
1156 /* -- Turn CSE back on. -- */
1157 set_optimize(rem_opt);
1158 current_ir_graph = rem;
1163 /********************************************************************/
1164 /* Apply inlineing to small methods. */
1165 /********************************************************************/
1167 static struct obstack temp_obst;
1169 /** Represents a possible inlinable call in a graph. */
1170 typedef struct _call_entry call_entry;
1171 struct _call_entry {
1172 ir_node *call; /**< the Call */
1173 ir_graph *callee; /**< the callee called here */
1174 call_entry *next; /**< for linking the next one */
1175 unsigned weight; /**< the weight of the call */
1179 * environment for inlining small irgs
1181 typedef struct _inline_env_t {
1182 struct obstack obst; /**< an obstack where call_entries are allocated on. */
1183 call_entry *head; /**< the head of the call entry list */
1184 call_entry *tail; /**< the tail of the call entry list */
1188 * Returns the irg called from a Call node. If the irg is not
1189 * known, NULL is returned.
1191 * @param call the call node
1193 static ir_graph *get_call_called_irg(ir_node *call) {
1196 addr = get_Call_ptr(call);
1197 if (is_Global(addr)) {
1198 ir_entity *ent = get_Global_entity(addr);
1199 return get_entity_irg(ent);
1206 * Walker: Collect all calls to known graphs inside a graph.
1208 static void collect_calls(ir_node *call, void *env) {
1209 if (is_Call(call)) {
1210 ir_graph *called_irg = get_call_called_irg(call);
1212 if (called_irg != NULL) {
1213 /* The Call node calls a locally defined method. Remember to inline. */
1214 inline_env_t *ienv = env;
1215 call_entry *entry = obstack_alloc(&ienv->obst, sizeof(*entry));
1217 entry->callee = called_irg;
1221 if (ienv->tail == NULL)
1224 ienv->tail->next = entry;
1231 * Inlines all small methods at call sites where the called address comes
1232 * from a Const node that references the entity representing the called
1234 * The size argument is a rough measure for the code size of the method:
1235 * Methods where the obstack containing the firm graph is smaller than
1238 void inline_small_irgs(ir_graph *irg, int size) {
1239 ir_graph *rem = current_ir_graph;
1243 current_ir_graph = irg;
1244 /* Handle graph state */
1245 assert(get_irg_phase_state(irg) != phase_building);
1246 free_callee_info(irg);
1248 /* Find Call nodes to inline.
1249 (We can not inline during a walk of the graph, as inlineing the same
1250 method several times changes the visited flag of the walked graph:
1251 after the first inlineing visited of the callee equals visited of
1252 the caller. With the next inlineing both are increased.) */
1253 obstack_init(&env.obst);
1254 env.head = env.tail = NULL;
1255 irg_walk_graph(irg, NULL, collect_calls, &env);
1257 if (env.head != NULL) {
1258 /* There are calls to inline */
1259 collect_phiprojs(irg);
1260 for (entry = env.head; entry != NULL; entry = entry->next) {
1261 ir_graph *callee = entry->callee;
1262 if (((_obstack_memory_used(callee->obst) - (int)obstack_room(callee->obst)) < size) ||
1263 (get_irg_inline_property(callee) >= irg_inline_forced)) {
1264 inline_method(entry->call, callee);
1268 obstack_free(&env.obst, NULL);
1269 current_ir_graph = rem;
1273 * Environment for inlining irgs.
1276 int n_nodes; /**< Number of nodes in graph except Id, Tuple, Proj, Start, End. */
1277 int n_blocks; /**< Number of Blocks in graph without Start and End block. */
1278 int n_nodes_orig; /**< for statistics */
1279 call_entry *call_head; /**< The head of the list of all call nodes in this graph. */
1280 call_entry *call_tail; /**< The tail of the list of all call nodes in this graph .*/
1281 int n_call_nodes; /**< Number of Call nodes in the graph. */
1282 int n_call_nodes_orig; /**< for statistics */
1283 int n_callers; /**< Number of known graphs that call this graphs. */
1284 int n_callers_orig; /**< for statistics */
1285 unsigned got_inline:1; /**< Set, if at least one call inside this graph was inlined. */
1286 unsigned local_vars:1; /**< Set, if a inlined function gets the address of an inlined variable. */
1287 unsigned *local_weights; /**< Once allocated, the beneficial weight for transmitting local addresses. */
1291 * Allocate a new environment for inlining.
1293 static inline_irg_env *alloc_inline_irg_env(void) {
1294 inline_irg_env *env = obstack_alloc(&temp_obst, sizeof(*env));
1295 env->n_nodes = -2; /* do not count count Start, End */
1296 env->n_blocks = -2; /* do not count count Start, End Block */
1297 env->n_nodes_orig = -2; /* do not count Start, End */
1298 env->call_head = NULL;
1299 env->call_tail = NULL;
1300 env->n_call_nodes = 0;
1301 env->n_call_nodes_orig = 0;
1303 env->n_callers_orig = 0;
1304 env->got_inline = 0;
1305 env->local_vars = 0;
1306 env->local_weights = NULL;
1310 typedef struct walker_env {
1311 inline_irg_env *x; /**< the inline environment */
1312 char ignore_runtime; /**< the ignore runtime flag */
1313 char ignore_callers; /**< if set, do change callers data */
1317 * post-walker: collect all calls in the inline-environment
1318 * of a graph and sum some statistics.
1320 static void collect_calls2(ir_node *call, void *ctx) {
1322 inline_irg_env *x = env->x;
1323 ir_opcode code = get_irn_opcode(call);
1327 /* count meaningful nodes in irg */
1328 if (code != iro_Proj && code != iro_Tuple && code != iro_Sync) {
1329 if (code != iro_Block) {
1337 if (code != iro_Call) return;
1339 /* check, if it's a runtime call */
1340 if (env->ignore_runtime) {
1341 ir_node *symc = get_Call_ptr(call);
1343 if (is_Global(symc)) {
1344 ir_entity *ent = get_Global_entity(symc);
1346 if (get_entity_additional_properties(ent) & mtp_property_runtime)
1351 /* collect all call nodes */
1353 ++x->n_call_nodes_orig;
1355 callee = get_call_called_irg(call);
1356 if (callee != NULL) {
1357 if (! env->ignore_callers) {
1358 inline_irg_env *callee_env = get_irg_link(callee);
1359 /* count all static callers */
1360 ++callee_env->n_callers;
1361 ++callee_env->n_callers_orig;
1364 /* link it in the list of possible inlinable entries */
1365 entry = obstack_alloc(&temp_obst, sizeof(*entry));
1367 entry->callee = callee;
1369 if (x->call_tail == NULL)
1370 x->call_head = entry;
1372 x->call_tail->next = entry;
1373 x->call_tail = entry;
1378 * Returns TRUE if the number of callers is 0 in the irg's environment,
1379 * hence this irg is a leave.
1381 INLINE static int is_leave(ir_graph *irg) {
1382 inline_irg_env *env = get_irg_link(irg);
1383 return env->n_call_nodes == 0;
1387 * Returns TRUE if the number of nodes in the callee is
1388 * smaller then size in the irg's environment.
1390 INLINE static int is_smaller(ir_graph *callee, int size) {
1391 inline_irg_env *env = get_irg_link(callee);
1392 return env->n_nodes < size;
1396 * Append the nodes of the list src to the nodes of the list in environment dst.
1398 static void append_call_list(inline_irg_env *dst, call_entry *src) {
1399 call_entry *entry, *nentry;
1401 /* Note that the src list points to Call nodes in the inlined graph, but
1402 we need Call nodes in our graph. Luckily the inliner leaves this information
1403 in the link field. */
1404 for (entry = src; entry != NULL; entry = entry->next) {
1405 nentry = obstack_alloc(&temp_obst, sizeof(*nentry));
1406 nentry->call = get_irn_link(entry->call);
1407 nentry->callee = entry->callee;
1408 nentry->next = NULL;
1409 dst->call_tail->next = nentry;
1410 dst->call_tail = nentry;
1415 * Inlines small leave methods at call sites where the called address comes
1416 * from a Const node that references the entity representing the called
1418 * The size argument is a rough measure for the code size of the method:
1419 * Methods where the obstack containing the firm graph is smaller than
1422 void inline_leave_functions(int maxsize, int leavesize, int size, int ignore_runtime) {
1423 inline_irg_env *env;
1429 call_entry *entry, *tail;
1430 const call_entry *centry;
1431 pmap *copied_graphs;
1432 pmap_entry *pm_entry;
1434 rem = current_ir_graph;
1435 obstack_init(&temp_obst);
1437 /* a map for the copied graphs, used to inline recursive calls */
1438 copied_graphs = pmap_create();
1440 /* extend all irgs by a temporary data structure for inlining. */
1441 n_irgs = get_irp_n_irgs();
1442 for (i = 0; i < n_irgs; ++i)
1443 set_irg_link(get_irp_irg(i), alloc_inline_irg_env());
1445 /* Precompute information in temporary data structure. */
1446 wenv.ignore_runtime = ignore_runtime;
1447 wenv.ignore_callers = 0;
1448 for (i = 0; i < n_irgs; ++i) {
1449 ir_graph *irg = get_irp_irg(i);
1451 assert(get_irg_phase_state(irg) != phase_building);
1452 free_callee_info(irg);
1454 wenv.x = get_irg_link(irg);
1455 irg_walk_graph(irg, NULL, collect_calls2, &wenv);
1458 /* -- and now inline. -- */
1460 /* Inline leaves recursively -- we might construct new leaves. */
1464 for (i = 0; i < n_irgs; ++i) {
1466 int phiproj_computed = 0;
1468 current_ir_graph = get_irp_irg(i);
1469 env = (inline_irg_env *)get_irg_link(current_ir_graph);
1472 for (entry = env->call_head; entry != NULL; entry = entry->next) {
1475 if (env->n_nodes > maxsize) break;
1478 callee = entry->callee;
1480 if (is_leave(callee) && (
1481 is_smaller(callee, leavesize) || (get_irg_inline_property(callee) >= irg_inline_forced))) {
1482 if (!phiproj_computed) {
1483 phiproj_computed = 1;
1484 collect_phiprojs(current_ir_graph);
1486 did_inline = inline_method(call, callee);
1489 inline_irg_env *callee_env = (inline_irg_env *)get_irg_link(callee);
1491 /* was inlined, must be recomputed */
1492 phiproj_computed = 0;
1494 /* Do some statistics */
1495 env->got_inline = 1;
1496 --env->n_call_nodes;
1497 env->n_nodes += callee_env->n_nodes;
1498 --callee_env->n_callers;
1500 /* remove this call from the list */
1502 tail->next = entry->next;
1504 env->call_head = entry->next;
1510 env->call_tail = tail;
1512 } while (did_inline);
1514 /* inline other small functions. */
1515 for (i = 0; i < n_irgs; ++i) {
1517 int phiproj_computed = 0;
1519 current_ir_graph = get_irp_irg(i);
1520 env = (inline_irg_env *)get_irg_link(current_ir_graph);
1522 /* note that the list of possible calls is updated during the process */
1524 for (entry = env->call_head; entry != NULL; entry = entry->next) {
1529 callee = entry->callee;
1531 e = pmap_find(copied_graphs, callee);
1534 * Remap callee if we have a copy.
1535 * FIXME: Should we do this only for recursive Calls ?
1540 if (((is_smaller(callee, size) && (env->n_nodes < maxsize)) || /* small function */
1541 (get_irg_inline_property(callee) >= irg_inline_forced))) {
1542 if (current_ir_graph == callee) {
1544 * Recursive call: we cannot directly inline because we cannot walk
1545 * the graph and change it. So we have to make a copy of the graph
1549 inline_irg_env *callee_env;
1553 * No copy yet, create one.
1554 * Note that recursive methods are never leaves, so it is sufficient
1555 * to test this condition here.
1557 copy = create_irg_copy(callee);
1559 /* create_irg_copy() destroys the Proj links, recompute them */
1560 phiproj_computed = 0;
1562 /* allocate new environment */
1563 callee_env = alloc_inline_irg_env();
1564 set_irg_link(copy, callee_env);
1566 wenv.x = callee_env;
1567 wenv.ignore_callers = 1;
1568 irg_walk_graph(copy, NULL, collect_calls2, &wenv);
1571 * Enter the entity of the original graph. This is needed
1572 * for inline_method(). However, note that ent->irg still points
1573 * to callee, NOT to copy.
1575 set_irg_entity(copy, get_irg_entity(callee));
1577 pmap_insert(copied_graphs, callee, copy);
1580 /* we have only one caller: the original graph */
1581 callee_env->n_callers = 1;
1582 callee_env->n_callers_orig = 1;
1584 if (! phiproj_computed) {
1585 phiproj_computed = 1;
1586 collect_phiprojs(current_ir_graph);
1588 did_inline = inline_method(call, callee);
1590 inline_irg_env *callee_env = (inline_irg_env *)get_irg_link(callee);
1592 /* was inlined, must be recomputed */
1593 phiproj_computed = 0;
1595 /* callee was inline. Append it's call list. */
1596 env->got_inline = 1;
1597 --env->n_call_nodes;
1598 append_call_list(env, callee_env->call_head);
1599 env->n_call_nodes += callee_env->n_call_nodes;
1600 env->n_nodes += callee_env->n_nodes;
1601 --callee_env->n_callers;
1603 /* after we have inlined callee, all called methods inside callee
1604 are now called once more */
1605 for (centry = callee_env->call_head; centry != NULL; centry = centry->next) {
1606 inline_irg_env *penv = get_irg_link(centry->callee);
1610 /* remove this call from the list */
1612 tail->next = entry->next;
1614 env->call_head = entry->next;
1620 env->call_tail = tail;
1623 for (i = 0; i < n_irgs; ++i) {
1624 irg = get_irp_irg(i);
1625 env = (inline_irg_env *)get_irg_link(irg);
1627 if (env->got_inline) {
1628 optimize_graph_df(irg);
1631 if (env->got_inline || (env->n_callers_orig != env->n_callers)) {
1632 DB((dbg, SET_LEVEL_1, "Nodes:%3d ->%3d, calls:%3d ->%3d, callers:%3d ->%3d, -- %s\n",
1633 env->n_nodes_orig, env->n_nodes, env->n_call_nodes_orig, env->n_call_nodes,
1634 env->n_callers_orig, env->n_callers,
1635 get_entity_name(get_irg_entity(irg))));
1639 /* kill the copied graphs: we don't need them anymore */
1640 foreach_pmap(copied_graphs, pm_entry) {
1641 ir_graph *copy = pm_entry->value;
1643 /* reset the entity, otherwise it will be deleted in the next step ... */
1644 set_irg_entity(copy, NULL);
1645 free_ir_graph(copy);
1647 pmap_destroy(copied_graphs);
1649 obstack_free(&temp_obst, NULL);
1650 current_ir_graph = rem;
1654 * Calculate the parameter weights for transmitting the address of a local variable.
1656 static unsigned calc_method_local_weight(ir_node *arg) {
1658 unsigned v, weight = 0;
1660 for (i = get_irn_n_outs(arg) - 1; i >= 0; --i) {
1661 ir_node *succ = get_irn_out(arg, i);
1663 switch (get_irn_opcode(succ)) {
1666 /* Loads and Store can be removed */
1670 /* check if all args are constant */
1671 for (j = get_Sel_n_indexs(succ) - 1; j >= 0; --j) {
1672 ir_node *idx = get_Sel_index(succ, j);
1673 if (! is_Const(idx))
1676 /* Check users on this Sel. Note: if a 0 is returned here, there was
1677 some unsupported node. */
1678 v = calc_method_local_weight(succ);
1681 /* we can kill one Sel with constant indexes, this is cheap */
1685 /* when looking backward we might find Id nodes */
1686 weight += calc_method_local_weight(succ);
1689 /* unoptimized tuple */
1690 for (j = get_Tuple_n_preds(succ) - 1; j >= 0; --j) {
1691 ir_node *pred = get_Tuple_pred(succ, j);
1693 /* look for Proj(j) */
1694 for (k = get_irn_n_outs(succ) - 1; k >= 0; --k) {
1695 ir_node *succ_succ = get_irn_out(succ, k);
1696 if (is_Proj(succ_succ)) {
1697 if (get_Proj_proj(succ_succ) == j) {
1699 weight += calc_method_local_weight(succ_succ);
1702 /* this should NOT happen */
1709 /* any other node: unsupported yet or bad. */
1717 * Calculate the parameter weights for transmitting the address of a local variable.
1719 static void analyze_irg_local_weights(inline_irg_env *env, ir_graph *irg) {
1720 ir_entity *ent = get_irg_entity(irg);
1722 int nparams, i, proj_nr;
1723 ir_node *irg_args, *arg;
1725 mtp = get_entity_type(ent);
1726 nparams = get_method_n_params(mtp);
1728 /* allocate a new array. currently used as 'analysed' flag */
1729 env->local_weights = NEW_ARR_D(unsigned, &temp_obst, nparams);
1731 /* If the method haven't parameters we have nothing to do. */
1735 assure_irg_outs(irg);
1736 irg_args = get_irg_args(irg);
1737 for (i = get_irn_n_outs(irg_args) - 1; i >= 0; --i) {
1738 arg = get_irn_out(irg_args, i);
1739 proj_nr = get_Proj_proj(arg);
1740 env->local_weights[proj_nr] = calc_method_local_weight(arg);
1745 * Calculate the benefice for transmitting an local variable address.
1746 * After inlining, the local variable might be transformed into a
1747 * SSA variable by scalar_replacement().
1749 static unsigned get_method_local_adress_weight(ir_graph *callee, int pos) {
1750 inline_irg_env *env = get_irg_link(callee);
1752 if (env->local_weights != NULL) {
1753 if (pos < ARR_LEN(env->local_weights))
1754 return env->local_weights[pos];
1758 analyze_irg_local_weights(env, callee);
1760 if (pos < ARR_LEN(env->local_weights))
1761 return env->local_weights[pos];
1766 * calculate a benefice value for inlining the given call.
1768 static int calc_inline_benefice(ir_node *call, ir_graph *callee, unsigned *local_adr) {
1769 ir_entity *ent = get_irg_entity(callee);
1776 inline_irg_env *curr_env, *callee_env;
1778 if (get_entity_additional_properties(ent) & mtp_property_noreturn) {
1779 /* do NOT inline noreturn calls */
1783 /* costs for every passed parameter */
1784 n_params = get_Call_n_params(call);
1785 mtp = get_entity_type(ent);
1786 cc = get_method_calling_convention(mtp);
1787 if (cc & cc_reg_param) {
1788 /* register parameter, smaller costs for register parameters */
1789 int max_regs = cc & ~cc_bits;
1791 if (max_regs < n_params)
1792 weight += max_regs * 2 + (n_params - max_regs) * 5;
1794 weight += n_params * 2;
1796 /* parameters are passed an stack */
1797 weight += 5 * n_params;
1800 /* constant parameters improve the benefice */
1801 frame_ptr = get_irg_frame(current_ir_graph);
1802 for (i = 0; i < n_params; ++i) {
1803 ir_node *param = get_Call_param(call, i);
1805 if (is_Const(param) || is_SymConst(param))
1806 weight += get_method_param_weight(ent, i);
1807 else if (is_Sel(param) && get_Sel_ptr(param) == frame_ptr) {
1809 * An address of a local variable is transmitted. After inlining,
1810 * scalar_replacement might be able to remove the local variable,
1813 v = get_method_local_adress_weight(callee, i);
1820 callee_env = get_irg_link(callee);
1821 if (get_entity_visibility(ent) == visibility_local &&
1822 callee_env->n_callers_orig == 1 &&
1823 callee != current_ir_graph) {
1824 /* we are the only caller, give big bonus */
1828 /* do not inline big functions */
1829 weight -= callee_env->n_nodes;
1831 /* reduce the benefice if the current function is already big */
1832 curr_env = get_irg_link(current_ir_graph);
1833 weight -= curr_env->n_nodes / 100;
1835 /* give a bonus for functions with one block */
1836 if (callee_env->n_blocks == 1)
1837 weight = weight * 3 / 2;
1843 * Heuristic inliner. Calculates a benifice value for every call and inlines
1844 * those calls with a value higher than the threshold.
1846 void inline_functions(int inline_threshold) {
1847 inline_irg_env *env;
1852 call_entry *entry, *tail;
1853 const call_entry *centry;
1854 pmap *copied_graphs;
1855 pmap_entry *pm_entry;
1857 rem = current_ir_graph;
1858 obstack_init(&temp_obst);
1860 /* a map for the copied graphs, used to inline recursive calls */
1861 copied_graphs = pmap_create();
1863 /* extend all irgs by a temporary data structure for inlining. */
1864 n_irgs = get_irp_n_irgs();
1865 for (i = 0; i < n_irgs; ++i)
1866 set_irg_link(get_irp_irg(i), alloc_inline_irg_env());
1868 /* Precompute information in temporary data structure. */
1869 wenv.ignore_runtime = 0;
1870 wenv.ignore_callers = 0;
1871 for (i = 0; i < n_irgs; ++i) {
1872 ir_graph *irg = get_irp_irg(i);
1874 assert(get_irg_phase_state(irg) != phase_building);
1875 free_callee_info(irg);
1877 wenv.x = get_irg_link(irg);
1878 irg_walk_graph(irg, NULL, collect_calls2, &wenv);
1881 /* -- and now inline. -- */
1882 for (i = 0; i < n_irgs; ++i) {
1883 int phiproj_computed = 0;
1885 ir_graph *irg = get_irp_irg(i);
1887 current_ir_graph = irg;
1888 env = get_irg_link(irg);
1890 /* note that the list of possible calls is updated during the process */
1892 for (entry = env->call_head; entry != NULL; entry = entry->next) {
1899 callee = entry->callee;
1901 /* calculate the benifice on the original call to prevent excessive inlining */
1903 benefice = calc_inline_benefice(call, callee, &local_adr);
1904 DB((dbg, SET_LEVEL_2, "In %+F Call %+F has benefice %d\n", irg, callee, benefice));
1906 e = pmap_find(copied_graphs, callee);
1909 * Remap callee if we have a copy.
1910 * FIXME: Should we do this only for recursive Calls ?
1915 if (benefice > -inline_threshold ||
1916 (get_irg_inline_property(callee) >= irg_inline_forced)) {
1917 if (current_ir_graph == callee) {
1919 * Recursive call: we cannot directly inline because we cannot walk
1920 * the graph and change it. So we have to make a copy of the graph
1924 inline_irg_env *callee_env;
1928 * No copy yet, create one.
1929 * Note that recursive methods are never leaves, so it is sufficient
1930 * to test this condition here.
1932 copy = create_irg_copy(callee);
1934 /* create_irg_copy() destroys the Proj links, recompute them */
1935 phiproj_computed = 0;
1937 /* allocate new environment */
1938 callee_env = alloc_inline_irg_env();
1939 set_irg_link(copy, callee_env);
1941 wenv.x = callee_env;
1942 wenv.ignore_callers = 1;
1943 irg_walk_graph(copy, NULL, collect_calls2, &wenv);
1946 * Enter the entity of the original graph. This is needed
1947 * for inline_method(). However, note that ent->irg still points
1948 * to callee, NOT to copy.
1950 set_irg_entity(copy, get_irg_entity(callee));
1952 pmap_insert(copied_graphs, callee, copy);
1955 /* we have only one caller: the original graph */
1956 callee_env->n_callers = 1;
1957 callee_env->n_callers_orig = 1;
1959 if (! phiproj_computed) {
1960 phiproj_computed = 1;
1961 collect_phiprojs(current_ir_graph);
1963 did_inline = inline_method(call, callee);
1965 inline_irg_env *callee_env = (inline_irg_env *)get_irg_link(callee);
1967 /* was inlined, must be recomputed */
1968 phiproj_computed = 0;
1970 /* callee was inline. Append it's call list. */
1971 env->got_inline = 1;
1973 env->local_vars = 1;
1974 --env->n_call_nodes;
1975 append_call_list(env, callee_env->call_head);
1976 env->n_call_nodes += callee_env->n_call_nodes;
1977 env->n_nodes += callee_env->n_nodes;
1978 --callee_env->n_callers;
1980 /* after we have inlined callee, all called methods inside callee
1981 are now called once more */
1982 for (centry = callee_env->call_head; centry != NULL; centry = centry->next) {
1983 inline_irg_env *penv = get_irg_link(centry->callee);
1987 /* remove this call from the list */
1989 tail->next = entry->next;
1991 env->call_head = entry->next;
1997 env->call_tail = tail;
1999 if (env->got_inline) {
2000 /* this irg got calls inlined: optimize it */
2002 /* scalar replacement does not work well with Tuple nodes, so optimize them away */
2003 optimize_graph_df(irg);
2005 if (env->local_vars) {
2006 if (scalar_replacement_opt(irg)) {
2007 optimize_graph_df(irg);
2013 if (env->got_inline || (env->n_callers_orig != env->n_callers)) {
2014 DB((dbg, SET_LEVEL_1, "Nodes:%3d ->%3d, calls:%3d ->%3d, callers:%3d ->%3d, -- %s\n",
2015 env->n_nodes_orig, env->n_nodes, env->n_call_nodes_orig, env->n_call_nodes,
2016 env->n_callers_orig, env->n_callers,
2017 get_entity_name(get_irg_entity(irg))));
2021 /* kill the copied graphs: we don't need them anymore */
2022 foreach_pmap(copied_graphs, pm_entry) {
2023 ir_graph *copy = pm_entry->value;
2025 /* reset the entity, otherwise it will be deleted in the next step ... */
2026 set_irg_entity(copy, NULL);
2027 free_ir_graph(copy);
2029 pmap_destroy(copied_graphs);
2031 obstack_free(&temp_obst, NULL);
2032 current_ir_graph = rem;
2035 void firm_init_inline(void) {
2036 FIRM_DBG_REGISTER(dbg, "firm.opt.inline");