3 * File name: ir/ir/irgopt.c
4 * Purpose: Optimizations for a whole ir graph, i.e., a procedure.
5 * Author: Christian Schaefer, Goetz Lindenmaier
6 * Modified by: Sebastian Felis
9 * Copyright: (c) 1998-2003 Universität Karlsruhe
10 * Licence: This file protected by GPL - GNU GENERAL PUBLIC LICENSE.
21 #include "irgraph_t.h"
34 #include "pdeq.h" /* Fuer code placement */
39 #include "irbackedge_t.h"
45 #include "iredges_t.h"
48 /* Defined in iropt.c */
49 pset *new_identities (void);
50 void del_identities (pset *value_table);
51 void add_identities (pset *value_table, ir_node *node);
53 /*------------------------------------------------------------------*/
54 /* apply optimizations of iropt to all nodes. */
55 /*------------------------------------------------------------------*/
57 static void init_link (ir_node *n, void *env) {
58 set_irn_link(n, NULL);
61 #if 0 /* Old version. Avoids Ids.
62 This is not necessary: we do a post walk, and get_irn_n
63 removes ids anyways. So it's much cheaper to call the
64 optimization less often and use the exchange() algorithm. */
66 optimize_in_place_wrapper (ir_node *n, void *env) {
68 ir_node *optimized, *old;
70 irn_arity = get_irn_arity(n);
71 for (i = 0; i < irn_arity; i++) {
72 /* get_irn_n skips Id nodes, so comparison old != optimized does not
73 show all optimizations. Therefore always set new predecessor. */
74 old = get_irn_intra_n(n, i);
75 optimized = optimize_in_place_2(old);
76 set_irn_n(n, i, optimized);
79 if (get_irn_op(n) == op_Block) {
80 optimized = optimize_in_place_2(n);
81 if (optimized != n) exchange (n, optimized);
86 optimize_in_place_wrapper (ir_node *n, void *env) {
87 ir_node *optimized = optimize_in_place_2(n);
88 if (optimized != n) exchange (n, optimized);
93 static INLINE void do_local_optimize(ir_node *n) {
94 /* Handle graph state */
95 assert(get_irg_phase_state(current_ir_graph) != phase_building);
97 if (get_opt_global_cse())
98 set_irg_pinned(current_ir_graph, op_pin_state_floats);
99 if (get_irg_outs_state(current_ir_graph) == outs_consistent)
100 set_irg_outs_inconsistent(current_ir_graph);
101 set_irg_doms_inconsistent(current_ir_graph);
102 set_irg_loopinfo_inconsistent(current_ir_graph);
104 /* Clean the value_table in irg for the CSE. */
105 del_identities(current_ir_graph->value_table);
106 current_ir_graph->value_table = new_identities();
108 /* walk over the graph */
109 irg_walk(n, init_link, optimize_in_place_wrapper, NULL);
112 void local_optimize_node(ir_node *n) {
113 ir_graph *rem = current_ir_graph;
114 current_ir_graph = get_irn_irg(n);
116 do_local_optimize(n);
118 current_ir_graph = rem;
122 * Block-Walker: uses dominance depth to mark dead blocks.
124 static void kill_dead_blocks(ir_node *block, void *env)
126 if (get_Block_dom_depth(block) < 0)
127 set_Block_dead(block);
131 local_optimize_graph (ir_graph *irg) {
132 ir_graph *rem = current_ir_graph;
133 current_ir_graph = irg;
135 if (get_irg_dom_state(current_ir_graph) == dom_consistent)
136 irg_block_walk_graph(irg, NULL, kill_dead_blocks, NULL);
138 do_local_optimize(irg->end);
140 current_ir_graph = rem;
144 /*------------------------------------------------------------------*/
145 /* Routines for dead node elimination / copying garbage collection */
146 /* of the obstack. */
147 /*------------------------------------------------------------------*/
150 * Remember the new node in the old node by using a field all nodes have.
152 #define set_new_node(oldn, newn) set_irn_link(oldn, newn)
155 * Get this new node, before the old node is forgotten.
157 #define get_new_node(oldn) get_irn_link(oldn)
160 * Check if a new node was set.
162 #define has_new_node(n) (get_new_node(n) != NULL)
165 * We use the block_visited flag to mark that we have computed the
166 * number of useful predecessors for this block.
167 * Further we encode the new arity in this flag in the old blocks.
168 * Remembering the arity is useful, as it saves a lot of pointer
169 * accesses. This function is called for all Phi and Block nodes
173 compute_new_arity(ir_node *b) {
174 int i, res, irn_arity;
177 irg_v = get_irg_block_visited(current_ir_graph);
178 block_v = get_Block_block_visited(b);
179 if (block_v >= irg_v) {
180 /* we computed the number of preds for this block and saved it in the
182 return block_v - irg_v;
184 /* compute the number of good predecessors */
185 res = irn_arity = get_irn_arity(b);
186 for (i = 0; i < irn_arity; i++)
187 if (get_irn_opcode(get_irn_n(b, i)) == iro_Bad) res--;
188 /* save it in the flag. */
189 set_Block_block_visited(b, irg_v + res);
195 * Copies the node to the new obstack. The Ins of the new node point to
196 * the predecessors on the old obstack. For block/phi nodes not all
197 * predecessors might be copied. n->link points to the new node.
198 * For Phi and Block nodes the function allocates in-arrays with an arity
199 * only for useful predecessors. The arity is determined by counting
200 * the non-bad predecessors of the block.
202 * @param n The node to be copied
203 * @param env if non-NULL, the node number attribute will be copied to the new node
205 * Note: Also used for loop unrolling.
207 static void copy_node(ir_node *n, void *env) {
210 ir_op *op = get_irn_op(n);
211 int copy_node_nr = env != NULL;
213 /* The end node looses it's flexible in array. This doesn't matter,
214 as dead node elimination builds End by hand, inlineing doesn't use
216 /* assert(op == op_End || ((_ARR_DESCR(n->in))->cookie != ARR_F_MAGIC)); */
219 /* node copied already */
221 } else if (op == op_Block) {
223 new_arity = compute_new_arity(n);
224 n->attr.block.graph_arr = NULL;
226 block = get_nodes_block(n);
228 new_arity = compute_new_arity(block);
230 new_arity = get_irn_arity(n);
233 nn = new_ir_node(get_irn_dbg_info(n),
240 /* Copy the attributes. These might point to additional data. If this
241 was allocated on the old obstack the pointers now are dangling. This
242 frees e.g. the memory of the graph_arr allocated in new_immBlock. */
243 copy_node_attr(n, nn);
244 new_backedge_info(nn);
248 /* for easier debugging, we want to copy the node numbers too */
249 nn->node_nr = n->node_nr;
254 hook_dead_node_elim_subst(current_ir_graph, n, nn);
258 * Copies new predecessors of old node to new node remembered in link.
259 * Spare the Bad predecessors of Phi and Block nodes.
262 copy_preds (ir_node *n, void *env) {
266 nn = get_new_node(n);
268 /* printf("\n old node: "); DDMSG2(n);
269 printf(" new node: "); DDMSG2(nn);
270 printf(" arities: old: %d, new: %d\n", get_irn_arity(n), get_irn_arity(nn)); */
273 /* Don't copy Bad nodes. */
275 irn_arity = get_irn_arity(n);
276 for (i = 0; i < irn_arity; i++)
277 if (! is_Bad(get_irn_n(n, i))) {
278 set_irn_n (nn, j, get_new_node(get_irn_n(n, i)));
279 /*if (is_backedge(n, i)) set_backedge(nn, j);*/
282 /* repair the block visited flag from above misuse. Repair it in both
283 graphs so that the old one can still be used. */
284 set_Block_block_visited(nn, 0);
285 set_Block_block_visited(n, 0);
286 /* Local optimization could not merge two subsequent blocks if
287 in array contained Bads. Now it's possible.
288 We don't call optimize_in_place as it requires
289 that the fields in ir_graph are set properly. */
290 if ((get_opt_control_flow_straightening()) &&
291 (get_Block_n_cfgpreds(nn) == 1) &&
292 (get_irn_op(get_Block_cfgpred(nn, 0)) == op_Jmp)) {
293 ir_node *old = get_nodes_block(get_Block_cfgpred(nn, 0));
295 /* Jmp jumps into the block it is in -- deal self cycle. */
296 assert(is_Bad(get_new_node(get_irg_bad(current_ir_graph))));
297 exchange(nn, get_new_node(get_irg_bad(current_ir_graph)));
302 } else if (get_irn_op(n) == op_Phi) {
303 /* Don't copy node if corresponding predecessor in block is Bad.
304 The Block itself should not be Bad. */
305 block = get_nodes_block(n);
306 set_irn_n (nn, -1, get_new_node(block));
308 irn_arity = get_irn_arity(n);
309 for (i = 0; i < irn_arity; i++)
310 if (! is_Bad(get_irn_n(block, i))) {
311 set_irn_n (nn, j, get_new_node(get_irn_n(n, i)));
312 /*if (is_backedge(n, i)) set_backedge(nn, j);*/
315 /* If the pre walker reached this Phi after the post walker visited the
316 block block_visited is > 0. */
317 set_Block_block_visited(get_nodes_block(n), 0);
318 /* Compacting the Phi's ins might generate Phis with only one
320 if (get_irn_arity(nn) == 1)
321 exchange(nn, get_irn_n(nn, 0));
323 irn_arity = get_irn_arity(n);
324 for (i = -1; i < irn_arity; i++)
325 set_irn_n (nn, i, get_new_node(get_irn_n(n, i)));
327 /* Now the new node is complete. We can add it to the hash table for CSE.
328 @@@ inlining aborts if we identify End. Why? */
329 if (get_irn_op(nn) != op_End)
330 add_identities (current_ir_graph->value_table, nn);
334 * Copies the graph recursively, compacts the keep-alives of the end node.
336 * @param irg the graph to be copied
337 * @param copy_node_nr If non-zero, the node number will be copied
339 static void copy_graph(ir_graph *irg, int copy_node_nr) {
340 ir_node *oe, *ne, *ob, *nb, *om, *nm; /* old end, new end, old bad, new bad, old NoMem, new NoMem */
341 ir_node *ka; /* keep alive */
344 oe = get_irg_end(irg);
345 /* copy the end node by hand, allocate dynamic in array! */
346 ne = new_ir_node(get_irn_dbg_info(oe),
353 /* Copy the attributes. Well, there might be some in the future... */
354 copy_node_attr(oe, ne);
355 set_new_node(oe, ne);
357 /* copy the Bad node */
358 ob = get_irg_bad(irg);
359 nb = new_ir_node(get_irn_dbg_info(ob),
366 set_new_node(ob, nb);
368 /* copy the NoMem node */
369 om = get_irg_no_mem(irg);
370 nm = new_ir_node(get_irn_dbg_info(om),
377 set_new_node(om, nm);
379 /* copy the live nodes */
380 irg_walk(get_nodes_block(oe), copy_node, copy_preds, INT_TO_PTR(copy_node_nr));
381 /* copy_preds for the end node ... */
382 set_nodes_block(ne, get_new_node(get_nodes_block(oe)));
384 /*- ... and now the keep alives. -*/
385 /* First pick the not marked block nodes and walk them. We must pick these
386 first as else we will oversee blocks reachable from Phis. */
387 irn_arity = get_irn_arity(oe);
388 for (i = 0; i < irn_arity; i++) {
389 ka = get_irn_intra_n(oe, i);
391 (get_irn_visited(ka) < get_irg_visited(irg))) {
392 /* We must keep the block alive and copy everything reachable */
393 set_irg_visited(irg, get_irg_visited(irg)-1);
394 irg_walk(ka, copy_node, copy_preds, INT_TO_PTR(copy_node_nr));
395 add_End_keepalive(ne, get_new_node(ka));
399 /* Now pick other nodes. Here we will keep all! */
400 irn_arity = get_irn_arity(oe);
401 for (i = 0; i < irn_arity; i++) {
402 ka = get_irn_intra_n(oe, i);
404 if (get_irn_visited(ka) < get_irg_visited(irg)) {
405 /* We didn't copy the node yet. */
406 set_irg_visited(irg, get_irg_visited(irg)-1);
407 irg_walk(ka, copy_node, copy_preds, INT_TO_PTR(copy_node_nr));
409 add_End_keepalive(ne, get_new_node(ka));
413 /* start block sometimes only reached after keep alives */
414 set_nodes_block(nb, get_new_node(get_nodes_block(ob)));
415 set_nodes_block(nm, get_new_node(get_nodes_block(om)));
419 * Copies the graph reachable from current_ir_graph->end to the obstack
420 * in current_ir_graph and fixes the environment.
421 * Then fixes the fields in current_ir_graph containing nodes of the
424 * @param copy_node_nr If non-zero, the node number will be copied
427 copy_graph_env (int copy_node_nr) {
428 ir_graph *irg = current_ir_graph;
429 ir_node *old_end, *n;
430 /* Not all nodes remembered in irg might be reachable
431 from the end node. Assure their link is set to NULL, so that
432 we can test whether new nodes have been computed. */
433 set_irn_link(get_irg_frame (irg), NULL);
434 set_irn_link(get_irg_globals (irg), NULL);
435 set_irn_link(get_irg_args (irg), NULL);
436 set_irn_link(get_irg_initial_mem(irg), NULL);
437 set_irn_link(get_irg_bad (irg), NULL);
438 set_irn_link(get_irg_no_mem (irg), NULL);
440 /* we use the block walk flag for removing Bads from Blocks ins. */
441 inc_irg_block_visited(irg);
444 copy_graph(irg, copy_node_nr);
446 /* fix the fields in irg */
447 old_end = get_irg_end(irg);
448 set_irg_end (irg, get_new_node(old_end));
449 set_irg_end_except (irg, get_irg_end(irg));
450 set_irg_end_reg (irg, get_irg_end(irg));
452 set_irg_end_block (irg, get_new_node(get_irg_end_block(irg)));
454 n = get_irg_frame(irg);
455 if (!has_new_node(n)) {
456 copy_node (n, INT_TO_PTR(copy_node_nr));
459 n = get_irg_globals(irg);
460 if (!has_new_node(n)) {
461 copy_node (n, INT_TO_PTR(copy_node_nr));
464 n = get_irg_initial_mem(irg);
465 if (!has_new_node(n)) {
466 copy_node (n, INT_TO_PTR(copy_node_nr));
469 n = get_irg_args(irg);
470 if (!has_new_node(n)) {
471 copy_node (n, INT_TO_PTR(copy_node_nr));
474 n = get_irg_bad(irg);
475 if (!has_new_node(n)) {
476 copy_node(n, INT_TO_PTR(copy_node_nr));
479 n = get_irg_no_mem(irg);
480 if (!has_new_node(n)) {
481 copy_node(n, INT_TO_PTR(copy_node_nr));
484 set_irg_start (irg, get_new_node(get_irg_start(irg)));
485 set_irg_start_block(irg, get_new_node(get_irg_start_block(irg)));
486 set_irg_frame (irg, get_new_node(get_irg_frame(irg)));
487 set_irg_globals (irg, get_new_node(get_irg_globals(irg)));
488 set_irg_initial_mem(irg, get_new_node(get_irg_initial_mem(irg)));
489 set_irg_args (irg, get_new_node(get_irg_args(irg)));
490 set_irg_bad (irg, get_new_node(get_irg_bad(irg)));
491 set_irg_no_mem (irg, get_new_node(get_irg_no_mem(irg)));
495 * Copies all reachable nodes to a new obstack. Removes bad inputs
496 * from block nodes and the corresponding inputs from Phi nodes.
497 * Merges single exit blocks with single entry blocks and removes
499 * Adds all new nodes to a new hash table for CSE. Does not
500 * perform CSE, so the hash table might contain common subexpressions.
503 dead_node_elimination(ir_graph *irg) {
505 int rem_ipview = get_interprocedural_view();
506 struct obstack *graveyard_obst = NULL;
507 struct obstack *rebirth_obst = NULL;
509 if (get_opt_optimize() && get_opt_dead_node_elimination()) {
510 assert(! edges_activated(irg) && "dead node elimination requires disabled edges");
512 /* inform statistics that we started a dead-node elimination run */
513 hook_dead_node_elim(irg, 1);
515 /* Remember external state of current_ir_graph. */
516 rem = current_ir_graph;
517 current_ir_graph = irg;
518 set_interprocedural_view(0);
520 assert(get_irg_phase_state(current_ir_graph) != phase_building);
522 /* Handle graph state */
523 free_callee_info(current_ir_graph);
524 free_irg_outs(current_ir_graph);
527 /* @@@ so far we loose loops when copying */
528 free_loop_information(current_ir_graph);
530 set_irg_doms_inconsistent(irg);
532 /* A quiet place, where the old obstack can rest in peace,
533 until it will be cremated. */
534 graveyard_obst = irg->obst;
536 /* A new obstack, where the reachable nodes will be copied to. */
537 rebirth_obst = xmalloc (sizeof(*rebirth_obst));
538 current_ir_graph->obst = rebirth_obst;
539 obstack_init (current_ir_graph->obst);
541 /* We also need a new hash table for cse */
542 del_identities (irg->value_table);
543 irg->value_table = new_identities ();
545 /* Copy the graph from the old to the new obstack */
548 /* Free memory from old unoptimized obstack */
549 obstack_free(graveyard_obst, 0); /* First empty the obstack ... */
550 xfree (graveyard_obst); /* ... then free it. */
552 /* inform statistics that the run is over */
553 hook_dead_node_elim(irg, 0);
555 current_ir_graph = rem;
556 set_interprocedural_view(rem_ipview);
561 * Relink bad predecessors of a block and store the old in array to the
562 * link field. This function is called by relink_bad_predecessors().
563 * The array of link field starts with the block operand at position 0.
564 * If block has bad predecessors, create a new in array without bad preds.
565 * Otherwise let in array untouched.
567 static void relink_bad_block_predecessors(ir_node *n, void *env) {
568 ir_node **new_in, *irn;
569 int i, new_irn_n, old_irn_arity, new_irn_arity = 0;
571 /* if link field of block is NULL, look for bad predecessors otherwise
572 this is already done */
573 if (get_irn_op(n) == op_Block &&
574 get_irn_link(n) == NULL) {
576 /* save old predecessors in link field (position 0 is the block operand)*/
577 set_irn_link(n, get_irn_in(n));
579 /* count predecessors without bad nodes */
580 old_irn_arity = get_irn_arity(n);
581 for (i = 0; i < old_irn_arity; i++)
582 if (!is_Bad(get_irn_n(n, i))) new_irn_arity++;
584 /* arity changing: set new predecessors without bad nodes */
585 if (new_irn_arity < old_irn_arity) {
586 /* Get new predecessor array. We do not resize the array, as we must
587 keep the old one to update Phis. */
588 new_in = NEW_ARR_D (ir_node *, current_ir_graph->obst, (new_irn_arity+1));
590 /* set new predecessors in array */
593 for (i = 0; i < old_irn_arity; i++) {
594 irn = get_irn_n(n, i);
596 new_in[new_irn_n] = irn;
597 is_backedge(n, i) ? set_backedge(n, new_irn_n-1) : set_not_backedge(n, new_irn_n-1);
601 //ARR_SETLEN(int, n->attr.block.backedge, new_irn_arity);
602 ARR_SHRINKLEN(n->attr.block.backedge, new_irn_arity);
605 } /* ir node has bad predecessors */
607 } /* Block is not relinked */
611 * Relinks Bad predecessors from Blocks and Phis called by walker
612 * remove_bad_predecesors(). If n is a Block, call
613 * relink_bad_block_redecessors(). If n is a Phi-node, call also the relinking
614 * function of Phi's Block. If this block has bad predecessors, relink preds
617 static void relink_bad_predecessors(ir_node *n, void *env) {
618 ir_node *block, **old_in;
619 int i, old_irn_arity, new_irn_arity;
621 /* relink bad predecessors of a block */
622 if (get_irn_op(n) == op_Block)
623 relink_bad_block_predecessors(n, env);
625 /* If Phi node relink its block and its predecessors */
626 if (get_irn_op(n) == op_Phi) {
628 /* Relink predecessors of phi's block */
629 block = get_nodes_block(n);
630 if (get_irn_link(block) == NULL)
631 relink_bad_block_predecessors(block, env);
633 old_in = (ir_node **)get_irn_link(block); /* Of Phi's Block */
634 old_irn_arity = ARR_LEN(old_in);
636 /* Relink Phi predecessors if count of predecessors changed */
637 if (old_irn_arity != ARR_LEN(get_irn_in(block))) {
638 /* set new predecessors in array
639 n->in[0] remains the same block */
641 for(i = 1; i < old_irn_arity; i++)
642 if (!is_Bad((ir_node *)old_in[i])) {
643 n->in[new_irn_arity] = n->in[i];
644 is_backedge(n, i) ? set_backedge(n, new_irn_arity) : set_not_backedge(n, new_irn_arity);
648 ARR_SETLEN(ir_node *, n->in, new_irn_arity);
649 ARR_SETLEN(int, n->attr.phi_backedge, new_irn_arity);
652 } /* n is a Phi node */
656 * Removes Bad Bad predecessors from Blocks and the corresponding
657 * inputs to Phi nodes as in dead_node_elimination but without
659 * On walking up set the link field to NULL, on walking down call
660 * relink_bad_predecessors() (This function stores the old in array
661 * to the link field and sets a new in array if arity of predecessors
664 void remove_bad_predecessors(ir_graph *irg) {
665 irg_walk_graph(irg, init_link, relink_bad_predecessors, NULL);
672 __)|_| | \_/ | \_/(/_ |_/\__|__
674 The following stuff implements a facility that automatically patches
675 registered ir_node pointers to the new node when a dead node elimination occurs.
678 struct _survive_dce_t {
682 hook_entry_t dead_node_elim;
683 hook_entry_t dead_node_elim_subst;
686 typedef struct _survive_dce_list_t {
687 struct _survive_dce_list_t *next;
689 } survive_dce_list_t;
691 static void dead_node_hook(void *context, ir_graph *irg, int start)
693 survive_dce_t *sd = context;
695 /* Create a new map before the dead node elimination is performed. */
697 sd->new_places = pmap_create_ex(pmap_count(sd->places));
700 /* Patch back all nodes if dead node elimination is over and something is to be done. */
702 pmap_destroy(sd->places);
703 sd->places = sd->new_places;
704 sd->new_places = NULL;
708 static void dead_node_subst_hook(void *context, ir_graph *irg, ir_node *old, ir_node *nw)
710 survive_dce_t *sd = context;
711 survive_dce_list_t *list = pmap_get(sd->places, old);
713 /* If the node is to be patched back, write the new address to all registered locations. */
715 survive_dce_list_t *p;
717 for(p = list; p; p = p->next)
720 pmap_insert(sd->new_places, nw, list);
725 * Make a new Survive DCE environment.
727 survive_dce_t *new_survive_dce(void)
729 survive_dce_t *res = xmalloc(sizeof(res[0]));
730 obstack_init(&res->obst);
731 res->places = pmap_create();
732 res->new_places = NULL;
734 res->dead_node_elim.hook._hook_dead_node_elim = dead_node_hook;
735 res->dead_node_elim.context = res;
736 res->dead_node_elim.next = NULL;
738 res->dead_node_elim_subst.hook._hook_dead_node_elim_subst = dead_node_subst_hook;
739 res->dead_node_elim_subst.context = res;
740 res->dead_node_elim_subst.next = NULL;
742 register_hook(hook_dead_node_elim, &res->dead_node_elim);
743 register_hook(hook_dead_node_elim_subst, &res->dead_node_elim_subst);
748 * Free a Survive DCE environment.
750 void free_survive_dce(survive_dce_t *sd)
752 obstack_free(&sd->obst, NULL);
753 pmap_destroy(sd->places);
754 unregister_hook(hook_dead_node_elim, &sd->dead_node_elim);
755 unregister_hook(hook_dead_node_elim_subst, &sd->dead_node_elim_subst);
760 * Register a node pointer to be patched upon DCE.
761 * When DCE occurs, the node pointer specified by @p place will be
762 * patched to the new address of the node it is pointing to.
764 * @param sd The Survive DCE environment.
765 * @param place The address of the node pointer.
767 void survive_dce_register_irn(survive_dce_t *sd, ir_node **place)
770 ir_node *irn = *place;
771 survive_dce_list_t *curr = pmap_get(sd->places, irn);
772 survive_dce_list_t *nw = obstack_alloc(&sd->obst, sizeof(nw));
777 pmap_insert(sd->places, irn, nw);
781 /*--------------------------------------------------------------------*/
782 /* Functionality for inlining */
783 /*--------------------------------------------------------------------*/
786 * Copy node for inlineing. Updates attributes that change when
787 * inlineing but not for dead node elimination.
789 * Copies the node by calling copy_node() and then updates the entity if
790 * it's a local one. env must be a pointer of the frame type of the
791 * inlined procedure. The new entities must be in the link field of
795 copy_node_inline (ir_node *n, void *env) {
797 ir_type *frame_tp = (ir_type *)env;
800 if (get_irn_op(n) == op_Sel) {
801 nn = get_new_node (n);
803 if (get_entity_owner(get_Sel_entity(n)) == frame_tp) {
804 set_Sel_entity(nn, get_entity_link(get_Sel_entity(n)));
806 } else if (get_irn_op(n) == op_Block) {
807 nn = get_new_node (n);
808 nn->attr.block.irg = current_ir_graph;
812 static void find_addr(ir_node *node, void *env)
814 if (get_irn_opcode(node) == iro_Proj) {
815 if (get_Proj_proj(node) == pn_Start_P_value_arg_base)
821 * currently, we cannot inline two cases:
822 * - call with compound arguments
823 * - graphs that take the address of a parameter
825 * check these conditions here
827 static int can_inline(ir_node *call, ir_graph *called_graph)
829 ir_type *call_type = get_Call_type(call);
830 int params, ress, i, res;
831 assert(is_Method_type(call_type));
833 params = get_method_n_params(call_type);
834 ress = get_method_n_ress(call_type);
837 for (i = 0; i < params; ++i) {
838 ir_type *p_type = get_method_param_type(call_type, i);
840 if (is_compound_type(p_type))
845 for (i = 0; i < ress; ++i) {
846 ir_type *r_type = get_method_res_type(call_type, i);
848 if (is_compound_type(r_type))
853 irg_walk_graph(called_graph, find_addr, NULL, &res);
858 int inline_method(ir_node *call, ir_graph *called_graph) {
860 ir_node *post_call, *post_bl;
862 ir_node *end, *end_bl;
866 int arity, n_ret, n_exc, n_res, i, j, rem_opt, irn_arity;
868 ir_type *called_frame;
869 irg_inline_property prop = get_irg_inline_property(called_graph);
871 if ( (prop != irg_inline_forced) &&
872 (!get_opt_optimize() || !get_opt_inline() || (prop == irg_inline_forbidden))) return 0;
874 /* Do not inline variadic functions. */
875 if (get_method_variadicity(get_entity_type(get_irg_entity(called_graph))) == variadicity_variadic)
878 assert(get_method_n_params(get_entity_type(get_irg_entity(called_graph))) ==
879 get_method_n_params(get_Call_type(call)));
882 * currently, we cannot inline two cases:
883 * - call with compound arguments
884 * - graphs that take the address of a parameter
886 if (! can_inline(call, called_graph))
889 /* -- Turn off optimizations, this can cause problems when allocating new nodes. -- */
890 rem_opt = get_opt_optimize();
893 /* Handle graph state */
894 assert(get_irg_phase_state(current_ir_graph) != phase_building);
895 assert(get_irg_pinned(current_ir_graph) == op_pin_state_pinned);
896 assert(get_irg_pinned(called_graph) == op_pin_state_pinned);
897 if (get_irg_outs_state(current_ir_graph) == outs_consistent)
898 set_irg_outs_inconsistent(current_ir_graph);
899 set_irg_loopinfo_inconsistent(current_ir_graph);
900 set_irg_callee_info_state(current_ir_graph, irg_callee_info_inconsistent);
902 /* -- Check preconditions -- */
903 assert(is_Call(call));
904 /* @@@ does not work for InterfaceIII.java after cgana
905 assert(get_Call_type(call) == get_entity_type(get_irg_entity(called_graph)));
906 assert(smaller_type(get_entity_type(get_irg_entity(called_graph)),
907 get_Call_type(call)));
909 assert(get_type_tpop(get_Call_type(call)) == type_method);
910 if (called_graph == current_ir_graph) {
911 set_optimize(rem_opt);
915 /* here we know we WILL inline, so inform the statistics */
916 hook_inline(call, called_graph);
918 /* -- Decide how to handle exception control flow: Is there a handler
919 for the Call node, or do we branch directly to End on an exception?
921 0 There is a handler.
923 2 Exception handling not represented in Firm. -- */
925 ir_node *proj, *Mproj = NULL, *Xproj = NULL;
926 for (proj = (ir_node *)get_irn_link(call); proj; proj = (ir_node *)get_irn_link(proj)) {
927 assert(get_irn_op(proj) == op_Proj);
928 if (get_Proj_proj(proj) == pn_Call_X_except) Xproj = proj;
929 if (get_Proj_proj(proj) == pn_Call_M_except) Mproj = proj;
931 if (Mproj) { assert(Xproj); exc_handling = 0; } /* Mproj */
932 else if (Xproj) { exc_handling = 1; } /* !Mproj && Xproj */
933 else { exc_handling = 2; } /* !Mproj && !Xproj */
938 the procedure and later replaces the Start node of the called graph.
939 Post_call is the old Call node and collects the results of the called
940 graph. Both will end up being a tuple. -- */
941 post_bl = get_nodes_block(call);
942 set_irg_current_block(current_ir_graph, post_bl);
943 /* XxMxPxP of Start + parameter of Call */
944 in[pn_Start_X_initial_exec] = new_Jmp();
945 in[pn_Start_M] = get_Call_mem(call);
946 in[pn_Start_P_frame_base] = get_irg_frame(current_ir_graph);
947 in[pn_Start_P_globals] = get_irg_globals(current_ir_graph);
948 in[pn_Start_T_args] = new_Tuple(get_Call_n_params(call), get_Call_param_arr(call));
949 /* in[pn_Start_P_value_arg_base] = ??? */
950 pre_call = new_Tuple(5, in);
954 The new block gets the ins of the old block, pre_call and all its
955 predecessors and all Phi nodes. -- */
956 part_block(pre_call);
958 /* -- Prepare state for dead node elimination -- */
959 /* Visited flags in calling irg must be >= flag in called irg.
960 Else walker and arity computation will not work. */
961 if (get_irg_visited(current_ir_graph) <= get_irg_visited(called_graph))
962 set_irg_visited(current_ir_graph, get_irg_visited(called_graph)+1);
963 if (get_irg_block_visited(current_ir_graph)< get_irg_block_visited(called_graph))
964 set_irg_block_visited(current_ir_graph, get_irg_block_visited(called_graph));
965 /* Set pre_call as new Start node in link field of the start node of
966 calling graph and pre_calls block as new block for the start block
968 Further mark these nodes so that they are not visited by the
970 set_irn_link(get_irg_start(called_graph), pre_call);
971 set_irn_visited(get_irg_start(called_graph), get_irg_visited(current_ir_graph));
972 set_irn_link(get_irg_start_block(called_graph), get_nodes_block(pre_call));
973 set_irn_visited(get_irg_start_block(called_graph), get_irg_visited(current_ir_graph));
974 set_irn_link(get_irg_bad(called_graph), get_irg_bad(current_ir_graph));
975 set_irn_visited(get_irg_bad(called_graph), get_irg_visited(current_ir_graph));
977 /* Initialize for compaction of in arrays */
978 inc_irg_block_visited(current_ir_graph);
980 /* -- Replicate local entities of the called_graph -- */
981 /* copy the entities. */
982 called_frame = get_irg_frame_type(called_graph);
983 for (i = 0; i < get_class_n_members(called_frame); i++) {
984 entity *new_ent, *old_ent;
985 old_ent = get_class_member(called_frame, i);
986 new_ent = copy_entity_own(old_ent, get_cur_frame_type());
987 set_entity_link(old_ent, new_ent);
990 /* visited is > than that of called graph. With this trick visited will
991 remain unchanged so that an outer walker, e.g., searching the call nodes
992 to inline, calling this inline will not visit the inlined nodes. */
993 set_irg_visited(current_ir_graph, get_irg_visited(current_ir_graph)-1);
995 /* -- Performing dead node elimination inlines the graph -- */
996 /* Copies the nodes to the obstack of current_ir_graph. Updates links to new
998 /* @@@ endless loops are not copied!! -- they should be, I think... */
999 irg_walk(get_irg_end(called_graph), copy_node_inline, copy_preds,
1000 get_irg_frame_type(called_graph));
1002 /* Repair called_graph */
1003 set_irg_visited(called_graph, get_irg_visited(current_ir_graph));
1004 set_irg_block_visited(called_graph, get_irg_block_visited(current_ir_graph));
1005 set_Block_block_visited(get_irg_start_block(called_graph), 0);
1007 /* -- Merge the end of the inlined procedure with the call site -- */
1008 /* We will turn the old Call node into a Tuple with the following
1011 0: Phi of all Memories of Return statements.
1012 1: Jmp from new Block that merges the control flow from all exception
1013 predecessors of the old end block.
1014 2: Tuple of all arguments.
1015 3: Phi of Exception memories.
1016 In case the old Call directly branches to End on an exception we don't
1017 need the block merging all exceptions nor the Phi of the exception
1021 /* -- Precompute some values -- */
1022 end_bl = get_new_node(get_irg_end_block(called_graph));
1023 end = get_new_node(get_irg_end(called_graph));
1024 arity = get_irn_arity(end_bl); /* arity = n_exc + n_ret */
1025 n_res = get_method_n_ress(get_Call_type(call));
1027 res_pred = xmalloc (n_res * sizeof(*res_pred));
1028 cf_pred = xmalloc (arity * sizeof(*res_pred));
1030 set_irg_current_block(current_ir_graph, post_bl); /* just to make sure */
1032 /* -- archive keepalives -- */
1033 irn_arity = get_irn_arity(end);
1034 for (i = 0; i < irn_arity; i++)
1035 add_End_keepalive(get_irg_end(current_ir_graph), get_irn_n(end, i));
1037 /* The new end node will die. We need not free as the in array is on the obstack:
1038 copy_node() only generated 'D' arrays. */
1040 /* -- Replace Return nodes by Jump nodes. -- */
1042 for (i = 0; i < arity; i++) {
1044 ret = get_irn_n(end_bl, i);
1045 if (is_Return(ret)) {
1046 cf_pred[n_ret] = new_r_Jmp(current_ir_graph, get_nodes_block(ret));
1050 set_irn_in(post_bl, n_ret, cf_pred);
1052 /* -- Build a Tuple for all results of the method.
1053 Add Phi node if there was more than one Return. -- */
1054 turn_into_tuple(post_call, 4);
1055 /* First the Memory-Phi */
1057 for (i = 0; i < arity; i++) {
1058 ret = get_irn_n(end_bl, i);
1059 if (is_Return(ret)) {
1060 cf_pred[n_ret] = get_Return_mem(ret);
1064 phi = new_Phi(n_ret, cf_pred, mode_M);
1065 set_Tuple_pred(call, pn_Call_M_regular, phi);
1066 /* Conserve Phi-list for further inlinings -- but might be optimized */
1067 if (get_nodes_block(phi) == post_bl) {
1068 set_irn_link(phi, get_irn_link(post_bl));
1069 set_irn_link(post_bl, phi);
1071 /* Now the real results */
1073 for (j = 0; j < n_res; j++) {
1075 for (i = 0; i < arity; i++) {
1076 ret = get_irn_n(end_bl, i);
1077 if (get_irn_op(ret) == op_Return) {
1078 cf_pred[n_ret] = get_Return_res(ret, j);
1083 phi = new_Phi(n_ret, cf_pred, get_irn_mode(cf_pred[0]));
1087 /* Conserve Phi-list for further inlinings -- but might be optimized */
1088 if (get_nodes_block(phi) == post_bl) {
1089 set_irn_link(phi, get_irn_link(post_bl));
1090 set_irn_link(post_bl, phi);
1093 set_Tuple_pred(call, pn_Call_T_result, new_Tuple(n_res, res_pred));
1095 set_Tuple_pred(call, pn_Call_T_result, new_Bad());
1097 /* Finally the exception control flow.
1098 We have two (three) possible situations:
1099 First if the Call branches to an exception handler: We need to add a Phi node to
1100 collect the memory containing the exception objects. Further we need
1101 to add another block to get a correct representation of this Phi. To
1102 this block we add a Jmp that resolves into the X output of the Call
1103 when the Call is turned into a tuple.
1104 Second the Call branches to End, the exception is not handled. Just
1105 add all inlined exception branches to the End node.
1106 Third: there is no Exception edge at all. Handle as case two. */
1107 if (exc_handling == 0) {
1109 for (i = 0; i < arity; i++) {
1111 ret = get_irn_n(end_bl, i);
1112 if (is_fragile_op(skip_Proj(ret)) || (get_irn_op(skip_Proj(ret)) == op_Raise)) {
1113 cf_pred[n_exc] = ret;
1118 new_Block(n_exc, cf_pred); /* watch it: current_block is changed! */
1119 set_Tuple_pred(call, pn_Call_X_except, new_Jmp());
1120 /* The Phi for the memories with the exception objects */
1122 for (i = 0; i < arity; i++) {
1124 ret = skip_Proj(get_irn_n(end_bl, i));
1126 cf_pred[n_exc] = new_r_Proj(current_ir_graph, get_nodes_block(ret), ret, mode_M, 3);
1128 } else if (is_fragile_op(ret)) {
1129 /* We rely that all cfops have the memory output at the same position. */
1130 cf_pred[n_exc] = new_r_Proj(current_ir_graph, get_nodes_block(ret), ret, mode_M, 0);
1132 } else if (get_irn_op(ret) == op_Raise) {
1133 cf_pred[n_exc] = new_r_Proj(current_ir_graph, get_nodes_block(ret), ret, mode_M, 1);
1137 set_Tuple_pred(call, pn_Call_M_except, new_Phi(n_exc, cf_pred, mode_M));
1139 set_Tuple_pred(call, pn_Call_X_except, new_Bad());
1140 set_Tuple_pred(call, pn_Call_M_except, new_Bad());
1143 ir_node *main_end_bl;
1144 int main_end_bl_arity;
1145 ir_node **end_preds;
1147 /* assert(exc_handling == 1 || no exceptions. ) */
1149 for (i = 0; i < arity; i++) {
1150 ir_node *ret = get_irn_n(end_bl, i);
1152 if (is_fragile_op(skip_Proj(ret)) || (get_irn_op(skip_Proj(ret)) == op_Raise)) {
1153 cf_pred[n_exc] = ret;
1157 main_end_bl = get_irg_end_block(current_ir_graph);
1158 main_end_bl_arity = get_irn_arity(main_end_bl);
1159 end_preds = xmalloc ((n_exc + main_end_bl_arity) * sizeof(*end_preds));
1161 for (i = 0; i < main_end_bl_arity; ++i)
1162 end_preds[i] = get_irn_n(main_end_bl, i);
1163 for (i = 0; i < n_exc; ++i)
1164 end_preds[main_end_bl_arity + i] = cf_pred[i];
1165 set_irn_in(main_end_bl, n_exc + main_end_bl_arity, end_preds);
1166 set_Tuple_pred(call, pn_Call_X_except, new_Bad());
1167 set_Tuple_pred(call, pn_Call_M_except, new_Bad());
1173 #if 0 /* old. now better, correcter, faster implementation. */
1175 /* -- If the exception control flow from the inlined Call directly
1176 branched to the end block we now have the following control
1177 flow predecessor pattern: ProjX -> Tuple -> Jmp. We must
1178 remove the Jmp along with it's empty block and add Jmp's
1179 predecessors as predecessors of this end block. No problem if
1180 there is no exception, because then branches Bad to End which
1182 @@@ can't we know this beforehand: by getting the Proj(1) from
1183 the Call link list and checking whether it goes to Proj. */
1184 /* find the problematic predecessor of the end block. */
1185 end_bl = get_irg_end_block(current_ir_graph);
1186 for (i = 0; i < get_Block_n_cfgpreds(end_bl); i++) {
1187 cf_op = get_Block_cfgpred(end_bl, i);
1188 if (get_irn_op(cf_op) == op_Proj) {
1189 cf_op = get_Proj_pred(cf_op);
1190 if ((get_irn_op(cf_op) == op_Tuple) && (cf_op == call)) {
1191 /* There are unoptimized tuples from inlineing before when no exc */
1192 assert(get_Proj_proj(get_Block_cfgpred(end_bl, i)) == pn_Call_X_except);
1193 cf_op = get_Tuple_pred(cf_op, pn_Call_X_except);
1194 assert(get_irn_op(cf_op) == op_Jmp);
1200 if (i < get_Block_n_cfgpreds(end_bl)) {
1201 bl = get_nodes_block(cf_op);
1202 arity = get_Block_n_cfgpreds(end_bl) + get_Block_n_cfgpreds(bl) - 1;
1203 cf_pred = xmalloc (arity * sizeof(*cf_pred));
1204 for (j = 0; j < i; j++)
1205 cf_pred[j] = get_Block_cfgpred(end_bl, j);
1206 for (j = j; j < i + get_Block_n_cfgpreds(bl); j++)
1207 cf_pred[j] = get_Block_cfgpred(bl, j-i);
1208 for (j = j; j < arity; j++)
1209 cf_pred[j] = get_Block_cfgpred(end_bl, j-get_Block_n_cfgpreds(bl) +1);
1210 set_irn_in(end_bl, arity, cf_pred);
1212 /* Remove the exception pred from post-call Tuple. */
1213 set_Tuple_pred(call, pn_Call_X_except, new_Bad());
1218 /* -- Turn CSE back on. -- */
1219 set_optimize(rem_opt);
1224 /********************************************************************/
1225 /* Apply inlineing to small methods. */
1226 /********************************************************************/
1228 /* It makes no sense to inline too many calls in one procedure. Anyways,
1229 I didn't get a version with NEW_ARR_F to run. */
1230 #define MAX_INLINE 1024
1233 * environment for inlining small irgs
1235 typedef struct _inline_env_t {
1237 ir_node *calls[MAX_INLINE];
1241 * Returns the irg called from a Call node. If the irg is not
1242 * known, NULL is returned.
1244 static ir_graph *get_call_called_irg(ir_node *call) {
1246 ir_graph *called_irg = NULL;
1248 assert(is_Call(call));
1250 addr = get_Call_ptr(call);
1251 if ((get_irn_op(addr) == op_SymConst) && (get_SymConst_kind (addr) == symconst_addr_ent)) {
1252 called_irg = get_entity_irg(get_SymConst_entity(addr));
1258 static void collect_calls(ir_node *call, void *env) {
1261 if (! is_Call(call)) return;
1263 addr = get_Call_ptr(call);
1265 if (get_irn_op(addr) == op_SymConst) {
1266 if (get_SymConst_kind(addr) == symconst_addr_ent) {
1267 ir_graph *called_irg = get_entity_irg(get_SymConst_entity(addr));
1268 inline_env_t *ienv = (inline_env_t *)env;
1269 if (called_irg && ienv->pos < MAX_INLINE) {
1270 /* The Call node calls a locally defined method. Remember to inline. */
1271 ienv->calls[ienv->pos++] = call;
1278 * Inlines all small methods at call sites where the called address comes
1279 * from a Const node that references the entity representing the called
1281 * The size argument is a rough measure for the code size of the method:
1282 * Methods where the obstack containing the firm graph is smaller than
1285 void inline_small_irgs(ir_graph *irg, int size) {
1287 ir_graph *rem = current_ir_graph;
1288 inline_env_t env /* = {0, NULL}*/;
1290 if (!(get_opt_optimize() && get_opt_inline())) return;
1292 current_ir_graph = irg;
1293 /* Handle graph state */
1294 assert(get_irg_phase_state(current_ir_graph) != phase_building);
1295 free_callee_info(current_ir_graph);
1297 /* Find Call nodes to inline.
1298 (We can not inline during a walk of the graph, as inlineing the same
1299 method several times changes the visited flag of the walked graph:
1300 after the first inlineing visited of the callee equals visited of
1301 the caller. With the next inlineing both are increased.) */
1303 irg_walk(get_irg_end(irg), NULL, collect_calls, &env);
1305 if ((env.pos > 0) && (env.pos < MAX_INLINE)) {
1306 /* There are calls to inline */
1307 collect_phiprojs(irg);
1308 for (i = 0; i < env.pos; i++) {
1310 callee = get_entity_irg(get_SymConst_entity(get_Call_ptr(env.calls[i])));
1311 if (((_obstack_memory_used(callee->obst) - (int)obstack_room(callee->obst)) < size) ||
1312 (get_irg_inline_property(callee) == irg_inline_forced)) {
1313 inline_method(env.calls[i], callee);
1318 current_ir_graph = rem;
1322 * Environment for inlining irgs.
1325 int n_nodes; /**< Nodes in graph except Id, Tuple, Proj, Start, End */
1326 int n_nodes_orig; /**< for statistics */
1327 eset *call_nodes; /**< All call nodes in this graph */
1329 int n_call_nodes_orig; /**< for statistics */
1330 int n_callers; /**< Number of known graphs that call this graphs. */
1331 int n_callers_orig; /**< for statistics */
1335 * Allocate a new environment for inlining.
1337 static inline_irg_env *new_inline_irg_env(void) {
1338 inline_irg_env *env = xmalloc(sizeof(*env));
1339 env->n_nodes = -2; /* do not count count Start, End */
1340 env->n_nodes_orig = -2; /* do not count Start, End */
1341 env->call_nodes = eset_create();
1342 env->n_call_nodes = 0;
1343 env->n_call_nodes_orig = 0;
1345 env->n_callers_orig = 0;
1350 * destroy an environment for inlining.
1352 static void free_inline_irg_env(inline_irg_env *env) {
1353 eset_destroy(env->call_nodes);
1358 * post-walker: collect all calls in the inline-environment
1359 * of a graph and sum some statistics.
1361 static void collect_calls2(ir_node *call, void *env) {
1362 inline_irg_env *x = (inline_irg_env *)env;
1363 ir_op *op = get_irn_op(call);
1366 /* count meaningful nodes in irg */
1367 if (op != op_Proj && op != op_Tuple && op != op_Sync) {
1372 if (op != op_Call) return;
1374 /* collect all call nodes */
1375 eset_insert(x->call_nodes, call);
1377 x->n_call_nodes_orig++;
1379 /* count all static callers */
1380 callee = get_call_called_irg(call);
1382 inline_irg_env *callee_env = get_irg_link(callee);
1383 callee_env->n_callers++;
1384 callee_env->n_callers_orig++;
1389 * Returns TRUE if the number of callers in 0 in the irg's environment,
1390 * hence this irg is a leave.
1392 INLINE static int is_leave(ir_graph *irg) {
1393 return (((inline_irg_env *)get_irg_link(irg))->n_call_nodes == 0);
1397 * Returns TRUE if the number of callers is smaller size in the irg's environment.
1399 INLINE static int is_smaller(ir_graph *callee, int size) {
1400 return (((inline_irg_env *)get_irg_link(callee))->n_nodes < size);
1405 * Inlines small leave methods at call sites where the called address comes
1406 * from a Const node that references the entity representing the called
1408 * The size argument is a rough measure for the code size of the method:
1409 * Methods where the obstack containing the firm graph is smaller than
1412 void inline_leave_functions(int maxsize, int leavesize, int size) {
1413 inline_irg_env *env;
1414 int i, n_irgs = get_irp_n_irgs();
1415 ir_graph *rem = current_ir_graph;
1418 if (!(get_opt_optimize() && get_opt_inline())) return;
1420 /* extend all irgs by a temporary data structure for inlining. */
1421 for (i = 0; i < n_irgs; ++i)
1422 set_irg_link(get_irp_irg(i), new_inline_irg_env());
1424 /* Precompute information in temporary data structure. */
1425 for (i = 0; i < n_irgs; ++i) {
1426 current_ir_graph = get_irp_irg(i);
1427 assert(get_irg_phase_state(current_ir_graph) != phase_building);
1428 free_callee_info(current_ir_graph);
1430 irg_walk(get_irg_end(current_ir_graph), NULL, collect_calls2,
1431 get_irg_link(current_ir_graph));
1434 /* -- and now inline. -- */
1436 /* Inline leaves recursively -- we might construct new leaves. */
1437 while (did_inline) {
1440 for (i = 0; i < n_irgs; ++i) {
1442 int phiproj_computed = 0;
1444 current_ir_graph = get_irp_irg(i);
1445 env = (inline_irg_env *)get_irg_link(current_ir_graph);
1447 for (call = eset_first(env->call_nodes); call; call = eset_next(env->call_nodes)) {
1450 if (get_irn_op(call) == op_Tuple) continue; /* We already have inlined this call. */
1451 callee = get_call_called_irg(call);
1453 if (env->n_nodes > maxsize) continue; // break;
1455 if (callee && (is_leave(callee) && is_smaller(callee, leavesize))) {
1456 if (!phiproj_computed) {
1457 phiproj_computed = 1;
1458 collect_phiprojs(current_ir_graph);
1460 did_inline = inline_method(call, callee);
1463 /* Do some statistics */
1464 inline_irg_env *callee_env = (inline_irg_env *)get_irg_link(callee);
1465 env->n_call_nodes --;
1466 env->n_nodes += callee_env->n_nodes;
1467 callee_env->n_callers--;
1474 /* inline other small functions. */
1475 for (i = 0; i < n_irgs; ++i) {
1478 int phiproj_computed = 0;
1480 current_ir_graph = get_irp_irg(i);
1481 env = (inline_irg_env *)get_irg_link(current_ir_graph);
1483 /* we can not walk and change a set, nor remove from it.
1485 walkset = env->call_nodes;
1486 env->call_nodes = eset_create();
1487 for (call = eset_first(walkset); call; call = eset_next(walkset)) {
1490 if (get_irn_op(call) == op_Tuple) continue; /* We already inlined. */
1491 callee = get_call_called_irg(call);
1494 ((is_smaller(callee, size) && (env->n_nodes < maxsize)) || /* small function */
1495 (get_irg_inline_property(callee) == irg_inline_forced))) {
1496 if (!phiproj_computed) {
1497 phiproj_computed = 1;
1498 collect_phiprojs(current_ir_graph);
1500 if (inline_method(call, callee)) {
1501 inline_irg_env *callee_env = (inline_irg_env *)get_irg_link(callee);
1502 env->n_call_nodes--;
1503 eset_insert_all(env->call_nodes, callee_env->call_nodes); /* @@@ ??? This are the wrong nodes !? Not the copied ones. */
1504 env->n_call_nodes += callee_env->n_call_nodes;
1505 env->n_nodes += callee_env->n_nodes;
1506 callee_env->n_callers--;
1509 eset_insert(env->call_nodes, call);
1512 eset_destroy(walkset);
1515 for (i = 0; i < n_irgs; ++i) {
1516 current_ir_graph = get_irp_irg(i);
1518 env = (inline_irg_env *)get_irg_link(current_ir_graph);
1519 if ((env->n_call_nodes_orig != env->n_call_nodes) ||
1520 (env->n_callers_orig != env->n_callers))
1521 printf("Nodes:%3d ->%3d, calls:%3d ->%3d, callers:%3d ->%3d, -- %s\n",
1522 env->n_nodes_orig, env->n_nodes, env->n_call_nodes_orig, env->n_call_nodes,
1523 env->n_callers_orig, env->n_callers,
1524 get_entity_name(get_irg_entity(current_ir_graph)));
1526 free_inline_irg_env((inline_irg_env *)get_irg_link(current_ir_graph));
1529 current_ir_graph = rem;
1532 /*******************************************************************/
1533 /* Code Placement. Pins all floating nodes to a block where they */
1534 /* will be executed only if needed. */
1535 /*******************************************************************/
1538 * Returns non-zero, is a block is not reachable from Start.
1540 * @param block the block to test
1543 is_Block_unreachable(ir_node *block) {
1544 return is_Block_dead(block) || get_Block_dom_depth(block) < 0;
1548 * Find the earliest correct block for N. --- Place N into the
1549 * same Block as its dominance-deepest Input.
1551 * We have to avoid calls to get_nodes_block() here
1552 * because the graph is floating.
1554 * move_out_of_loops() expects that place_floats_early() have placed
1555 * all "living" nodes into a living block. That's why we must
1556 * move nodes in dead block with "live" successors into a valid
1558 * We move them just into the same block as it's successor (or
1559 * in case of a Phi into the effective use block). For Phi successors,
1560 * this may still be a dead block, but then there is no real use, as
1561 * the control flow will be dead later.
1564 place_floats_early(ir_node *n, pdeq *worklist)
1568 /* we must not run into an infinite loop */
1569 assert(irn_not_visited(n));
1570 mark_irn_visited(n);
1572 /* Place floating nodes. */
1573 if (get_irn_pinned(n) == op_pin_state_floats) {
1574 ir_node *curr_block = get_irn_n(n, -1);
1575 int in_dead_block = is_Block_unreachable(curr_block);
1577 ir_node *b = NULL; /* The block to place this node in */
1579 assert(get_irn_op(n) != op_Block);
1581 if ((get_irn_op(n) == op_Const) ||
1582 (get_irn_op(n) == op_SymConst) ||
1584 (get_irn_op(n) == op_Unknown)) {
1585 /* These nodes will not be placed by the loop below. */
1586 b = get_irg_start_block(current_ir_graph);
1590 /* find the block for this node. */
1591 irn_arity = get_irn_arity(n);
1592 for (i = 0; i < irn_arity; i++) {
1593 ir_node *pred = get_irn_n(n, i);
1594 ir_node *pred_block;
1596 if ((irn_not_visited(pred))
1597 && (get_irn_pinned(pred) == op_pin_state_floats)) {
1600 * If the current node is NOT in a dead block, but one of its
1601 * predecessors is, we must move the predecessor to a live block.
1602 * Such thing can happen, if global CSE chose a node from a dead block.
1603 * We move it simple to our block.
1604 * Note that neither Phi nor End nodes are floating, so we don't
1605 * need to handle them here.
1607 if (! in_dead_block) {
1608 if (get_irn_pinned(pred) == op_pin_state_floats &&
1609 is_Block_unreachable(get_irn_n(pred, -1)))
1610 set_nodes_block(pred, curr_block);
1612 place_floats_early(pred, worklist);
1616 * A node in the Bad block must stay in the bad block,
1617 * so don't compute a new block for it.
1622 /* Because all loops contain at least one op_pin_state_pinned node, now all
1623 our inputs are either op_pin_state_pinned or place_early() has already
1624 been finished on them. We do not have any unfinished inputs! */
1625 pred_block = get_irn_n(pred, -1);
1626 if ((!is_Block_dead(pred_block)) &&
1627 (get_Block_dom_depth(pred_block) > depth)) {
1629 depth = get_Block_dom_depth(pred_block);
1631 /* Avoid that the node is placed in the Start block */
1632 if ((depth == 1) && (get_Block_dom_depth(get_irn_n(n, -1)) > 1)) {
1633 b = get_Block_cfg_out(get_irg_start_block(current_ir_graph), 0);
1634 assert(b != get_irg_start_block(current_ir_graph));
1639 set_nodes_block(n, b);
1643 * Add predecessors of non floating nodes and non-floating predecessors
1644 * of floating nodes to worklist and fix their blocks if the are in dead block.
1646 irn_arity = get_irn_arity(n);
1648 if (get_irn_op(n) == op_End) {
1650 * Simplest case: End node. Predecessors are keep-alives,
1651 * no need to move out of dead block.
1653 for (i = -1; i < irn_arity; ++i) {
1654 ir_node *pred = get_irn_n(n, i);
1655 if (irn_not_visited(pred))
1656 pdeq_putr(worklist, pred);
1659 else if (is_Block(n)) {
1661 * Blocks: Predecessors are control flow, no need to move
1662 * them out of dead block.
1664 for (i = irn_arity - 1; i >= 0; --i) {
1665 ir_node *pred = get_irn_n(n, i);
1666 if (irn_not_visited(pred))
1667 pdeq_putr(worklist, pred);
1670 else if (is_Phi(n)) {
1672 ir_node *curr_block = get_irn_n(n, -1);
1673 int in_dead_block = is_Block_unreachable(curr_block);
1676 * Phi nodes: move nodes from dead blocks into the effective use
1677 * of the Phi-input if the Phi is not in a bad block.
1679 pred = get_irn_n(n, -1);
1680 if (irn_not_visited(pred))
1681 pdeq_putr(worklist, pred);
1683 for (i = irn_arity - 1; i >= 0; --i) {
1684 ir_node *pred = get_irn_n(n, i);
1686 if (irn_not_visited(pred)) {
1687 if (! in_dead_block &&
1688 get_irn_pinned(pred) == op_pin_state_floats &&
1689 is_Block_unreachable(get_irn_n(pred, -1))) {
1690 set_nodes_block(pred, get_Block_cfgpred_block(curr_block, i));
1692 pdeq_putr(worklist, pred);
1698 ir_node *curr_block = get_irn_n(n, -1);
1699 int in_dead_block = is_Block_unreachable(curr_block);
1702 * All other nodes: move nodes from dead blocks into the same block.
1704 pred = get_irn_n(n, -1);
1705 if (irn_not_visited(pred))
1706 pdeq_putr(worklist, pred);
1708 for (i = irn_arity - 1; i >= 0; --i) {
1709 ir_node *pred = get_irn_n(n, i);
1711 if (irn_not_visited(pred)) {
1712 if (! in_dead_block &&
1713 get_irn_pinned(pred) == op_pin_state_floats &&
1714 is_Block_unreachable(get_irn_n(pred, -1))) {
1715 set_nodes_block(pred, curr_block);
1717 pdeq_putr(worklist, pred);
1724 * Floating nodes form subgraphs that begin at nodes as Const, Load,
1725 * Start, Call and that end at op_pin_state_pinned nodes as Store, Call. Place_early
1726 * places all floating nodes reachable from its argument through floating
1727 * nodes and adds all beginnings at op_pin_state_pinned nodes to the worklist.
1729 static INLINE void place_early(pdeq *worklist) {
1731 inc_irg_visited(current_ir_graph);
1733 /* this inits the worklist */
1734 place_floats_early(get_irg_end(current_ir_graph), worklist);
1736 /* Work the content of the worklist. */
1737 while (!pdeq_empty(worklist)) {
1738 ir_node *n = pdeq_getl(worklist);
1739 if (irn_not_visited(n))
1740 place_floats_early(n, worklist);
1743 set_irg_outs_inconsistent(current_ir_graph);
1744 set_irg_pinned(current_ir_graph, op_pin_state_pinned);
1748 * Compute the deepest common ancestor of block and dca.
1750 static ir_node *calc_dca(ir_node *dca, ir_node *block)
1754 /* we do not want to place nodes in dead blocks */
1755 if (is_Block_dead(block))
1758 /* We found a first legal placement. */
1759 if (!dca) return block;
1761 /* Find a placement that is dominates both, dca and block. */
1762 while (get_Block_dom_depth(block) > get_Block_dom_depth(dca))
1763 block = get_Block_idom(block);
1765 while (get_Block_dom_depth(dca) > get_Block_dom_depth(block)) {
1766 dca = get_Block_idom(dca);
1769 while (block != dca)
1770 { block = get_Block_idom(block); dca = get_Block_idom(dca); }
1775 /** Deepest common dominance ancestor of DCA and CONSUMER of PRODUCER.
1776 * I.e., DCA is the block where we might place PRODUCER.
1777 * A data flow edge points from producer to consumer.
1780 consumer_dom_dca(ir_node *dca, ir_node *consumer, ir_node *producer)
1782 ir_node *block = NULL;
1784 /* Compute the latest block into which we can place a node so that it is
1786 if (get_irn_op(consumer) == op_Phi) {
1787 /* our consumer is a Phi-node, the effective use is in all those
1788 blocks through which the Phi-node reaches producer */
1790 ir_node *phi_block = get_nodes_block(consumer);
1791 irn_arity = get_irn_arity(consumer);
1793 for (i = 0; i < irn_arity; i++) {
1794 if (get_irn_n(consumer, i) == producer) {
1795 ir_node *new_block = get_nodes_block(get_Block_cfgpred(phi_block, i));
1797 if (! is_Block_unreachable(new_block))
1798 block = calc_dca(block, new_block);
1803 block = get_irn_n(producer, -1);
1806 assert(is_no_Block(consumer));
1807 block = get_nodes_block(consumer);
1810 /* Compute the deepest common ancestor of block and dca. */
1811 return calc_dca(dca, block);
1814 /* FIXME: the name clashes here with the function from ana/field_temperature.c
1816 static INLINE int get_irn_loop_depth(ir_node *n) {
1817 return get_loop_depth(get_irn_loop(n));
1821 * Move n to a block with less loop depth than it's current block. The
1822 * new block must be dominated by early.
1824 * @param n the node that should be moved
1825 * @param early the earliest block we can n move to
1828 move_out_of_loops (ir_node *n, ir_node *early)
1830 ir_node *best, *dca;
1834 /* Find the region deepest in the dominator tree dominating
1835 dca with the least loop nesting depth, but still dominated
1836 by our early placement. */
1837 dca = get_nodes_block(n);
1840 while (dca != early) {
1841 dca = get_Block_idom(dca);
1842 if (!dca || is_Bad(dca)) break; /* may be Bad if not reachable from Start */
1843 if (get_irn_loop_depth(dca) < get_irn_loop_depth(best)) {
1847 if (best != get_nodes_block(n)) {
1849 printf("Moving out of loop: "); DDMN(n);
1850 printf(" Outermost block: "); DDMN(early);
1851 printf(" Best block: "); DDMN(best);
1852 printf(" Innermost block: "); DDMN(get_nodes_block(n));
1854 set_nodes_block(n, best);
1859 * Find the latest legal block for N and place N into the
1860 * `optimal' Block between the latest and earliest legal block.
1861 * The `optimal' block is the dominance-deepest block of those
1862 * with the least loop-nesting-depth. This places N out of as many
1863 * loops as possible and then makes it as control dependent as
1867 place_floats_late(ir_node *n, pdeq *worklist)
1872 assert(irn_not_visited(n)); /* no multiple placement */
1874 mark_irn_visited(n);
1876 /* no need to place block nodes, control nodes are already placed. */
1877 if ((get_irn_op(n) != op_Block) &&
1879 (get_irn_mode(n) != mode_X)) {
1880 /* Remember the early_blk placement of this block to move it
1881 out of loop no further than the early_blk placement. */
1882 early_blk = get_irn_n(n, -1);
1885 * BEWARE: Here we also get code, that is live, but
1886 * was in a dead block. If the node is life, but because
1887 * of CSE in a dead block, we still might need it.
1890 /* Assure that our users are all placed, except the Phi-nodes.
1891 --- Each data flow cycle contains at least one Phi-node. We
1892 have to break the `user has to be placed before the
1893 producer' dependence cycle and the Phi-nodes are the
1894 place to do so, because we need to base our placement on the
1895 final region of our users, which is OK with Phi-nodes, as they
1896 are op_pin_state_pinned, and they never have to be placed after a
1897 producer of one of their inputs in the same block anyway. */
1898 for (i = get_irn_n_outs(n) - 1; i >= 0; --i) {
1899 ir_node *succ = get_irn_out(n, i);
1900 if (irn_not_visited(succ) && (get_irn_op(succ) != op_Phi))
1901 place_floats_late(succ, worklist);
1904 if (! is_Block_dead(early_blk)) {
1905 /* do only move things that where not dead */
1907 /* We have to determine the final block of this node... except for
1909 if ((get_irn_pinned(n) == op_pin_state_floats) &&
1910 (get_irn_op(n) != op_Const) &&
1911 (get_irn_op(n) != op_SymConst)) {
1912 ir_node *dca = NULL; /* deepest common ancestor in the
1913 dominator tree of all nodes'
1914 blocks depending on us; our final
1915 placement has to dominate DCA. */
1916 for (i = get_irn_n_outs(n) - 1; i >= 0; --i) {
1917 ir_node *succ = get_irn_out(n, i);
1920 if (get_irn_op(succ) == op_End) {
1922 * This consumer is the End node, a keep alive edge.
1923 * This is not a real consumer, so we ignore it
1928 /* ignore if succ is in dead code */
1929 succ_blk = get_irn_n(succ, -1);
1930 if (is_Block_unreachable(succ_blk))
1932 dca = consumer_dom_dca(dca, succ, n);
1935 set_nodes_block(n, dca);
1936 move_out_of_loops(n, early_blk);
1942 /* Add predecessors of all non-floating nodes on list. (Those of floating
1943 nodes are placed already and therefore are marked.) */
1944 for (i = 0; i < get_irn_n_outs(n); i++) {
1945 ir_node *succ = get_irn_out(n, i);
1946 if (irn_not_visited(get_irn_out(n, i))) {
1947 pdeq_putr(worklist, succ);
1952 static INLINE void place_late(pdeq *worklist) {
1954 inc_irg_visited(current_ir_graph);
1956 /* This fills the worklist initially. */
1957 place_floats_late(get_irg_start_block(current_ir_graph), worklist);
1959 /* And now empty the worklist again... */
1960 while (!pdeq_empty(worklist)) {
1961 ir_node *n = pdeq_getl(worklist);
1962 if (irn_not_visited(n))
1963 place_floats_late(n, worklist);
1967 void place_code(ir_graph *irg) {
1969 ir_graph *rem = current_ir_graph;
1971 current_ir_graph = irg;
1973 if (!(get_opt_optimize() && get_opt_global_cse())) return;
1975 /* Handle graph state */
1976 assert(get_irg_phase_state(irg) != phase_building);
1977 if (get_irg_dom_state(irg) != dom_consistent)
1980 if (1 || get_irg_loopinfo_state(irg) != loopinfo_consistent) {
1981 free_loop_information(irg);
1982 construct_backedges(irg);
1985 /* Place all floating nodes as early as possible. This guarantees
1986 a legal code placement. */
1987 worklist = new_pdeq();
1988 place_early(worklist);
1990 /* place_early() invalidates the outs, place_late needs them. */
1991 compute_irg_outs(irg);
1993 /* Now move the nodes down in the dominator tree. This reduces the
1994 unnecessary executions of the node. */
1995 place_late(worklist);
1997 set_irg_outs_inconsistent(current_ir_graph);
1998 set_irg_loopinfo_inconsistent(current_ir_graph);
2000 current_ir_graph = rem;
2004 * Called by walker of remove_critical_cf_edges().
2006 * Place an empty block to an edge between a blocks of multiple
2007 * predecessors and a block of multiple successors.
2010 * @param env Environment of walker. This field is unused and has
2013 static void walk_critical_cf_edges(ir_node *n, void *env) {
2015 ir_node *pre, *block, **in, *jmp;
2017 /* Block has multiple predecessors */
2018 if ((op_Block == get_irn_op(n)) &&
2019 (get_irn_arity(n) > 1)) {
2020 arity = get_irn_arity(n);
2022 if (n == get_irg_end_block(current_ir_graph))
2023 return; /* No use to add a block here. */
2025 for (i=0; i<arity; i++) {
2026 pre = get_irn_n(n, i);
2027 /* Predecessor has multiple successors. Insert new flow edge */
2028 if ((NULL != pre) &&
2029 (op_Proj == get_irn_op(pre)) &&
2030 op_Raise != get_irn_op(skip_Proj(pre))) {
2032 /* set predecessor array for new block */
2033 in = NEW_ARR_D (ir_node *, current_ir_graph->obst, 1);
2034 /* set predecessor of new block */
2036 block = new_Block(1, in);
2037 /* insert new jmp node to new block */
2038 set_cur_block(block);
2041 /* set successor of new block */
2042 set_irn_n(n, i, jmp);
2044 } /* predecessor has multiple successors */
2045 } /* for all predecessors */
2046 } /* n is a block */
2049 void remove_critical_cf_edges(ir_graph *irg) {
2050 if (get_opt_critical_edges())
2051 irg_walk_graph(irg, NULL, walk_critical_cf_edges, NULL);