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 Code Placement. Pins all floating nodes to a block where they
23 * will be executed only if needed.
24 * @author Christian Schaefer, Goetz Lindenmaier, Sebastian Felis,
38 * Returns non-zero, is a block is not reachable from Start.
40 * @param block the block to test
43 is_Block_unreachable(ir_node *block) {
44 return is_Block_dead(block) || get_Block_dom_depth(block) < 0;
48 * Find the earliest correct block for node n. --- Place n into the
49 * same Block as its dominance-deepest Input.
51 * We have to avoid calls to get_nodes_block() here
52 * because the graph is floating.
54 * move_out_of_loops() expects that place_floats_early() have placed
55 * all "living" nodes into a living block. That's why we must
56 * move nodes in dead block with "live" successors into a valid
58 * We move them just into the same block as it's successor (or
59 * in case of a Phi into the effective use block). For Phi successors,
60 * this may still be a dead block, but then there is no real use, as
61 * the control flow will be dead later.
63 * @param n the node to be placed
64 * @param worklist a worklist, predecessors of non-floating nodes are placed here
67 place_floats_early(ir_node *n, waitq *worklist) {
70 /* we must not run into an infinite loop */
71 assert(irn_not_visited(n));
74 /* Place floating nodes. */
75 if (get_irn_pinned(n) == op_pin_state_floats) {
76 ir_node *curr_block = get_nodes_block(n);
77 int in_dead_block = is_Block_unreachable(curr_block);
79 ir_node *b = NULL; /* The block to place this node in */
81 assert(is_no_Block(n));
83 if (is_irn_start_block_placed(n)) {
84 /* These nodes will not be placed by the loop below. */
85 b = get_irg_start_block(current_ir_graph);
89 /* find the block for this node. */
90 irn_arity = get_irn_arity(n);
91 for (i = 0; i < irn_arity; i++) {
92 ir_node *pred = get_irn_n(n, i);
95 if ((irn_not_visited(pred))
96 && (get_irn_pinned(pred) == op_pin_state_floats)) {
99 * If the current node is NOT in a dead block, but one of its
100 * predecessors is, we must move the predecessor to a live block.
101 * Such thing can happen, if global CSE chose a node from a dead block.
102 * We move it simply to our block.
103 * Note that neither Phi nor End nodes are floating, so we don't
104 * need to handle them here.
106 if (! in_dead_block) {
107 if (get_irn_pinned(pred) == op_pin_state_floats &&
108 is_Block_unreachable(get_nodes_block(pred)))
109 set_nodes_block(pred, curr_block);
111 place_floats_early(pred, worklist);
115 * A node in the Bad block must stay in the bad block,
116 * so don't compute a new block for it.
121 /* Because all loops contain at least one op_pin_state_pinned node, now all
122 our inputs are either op_pin_state_pinned or place_early() has already
123 been finished on them. We do not have any unfinished inputs! */
124 pred_block = get_nodes_block(pred);
125 if ((!is_Block_dead(pred_block)) &&
126 (get_Block_dom_depth(pred_block) > depth)) {
128 depth = get_Block_dom_depth(pred_block);
130 /* Avoid that the node is placed in the Start block if we are not in the
133 get_Block_dom_depth(get_nodes_block(n)) > 1 &&
134 get_irg_phase_state(current_ir_graph) != phase_backend) {
135 b = get_Block_cfg_out(get_irg_start_block(current_ir_graph), 0);
136 assert(b != get_irg_start_block(current_ir_graph));
141 set_nodes_block(n, b);
145 * Add predecessors of non floating nodes and non-floating predecessors
146 * of floating nodes to worklist and fix their blocks if the are in dead block.
148 irn_arity = get_irn_arity(n);
152 * Simplest case: End node. Predecessors are keep-alives,
153 * no need to move out of dead block.
155 for (i = -1; i < irn_arity; ++i) {
156 ir_node *pred = get_irn_n(n, i);
157 if (irn_not_visited(pred))
158 waitq_put(worklist, pred);
160 } else if (is_Block(n)) {
162 * Blocks: Predecessors are control flow, no need to move
163 * them out of dead block.
165 for (i = irn_arity - 1; i >= 0; --i) {
166 ir_node *pred = get_irn_n(n, i);
167 if (irn_not_visited(pred))
168 waitq_put(worklist, pred);
170 } else if (is_Phi(n)) {
172 ir_node *curr_block = get_nodes_block(n);
173 int in_dead_block = is_Block_unreachable(curr_block);
176 * Phi nodes: move nodes from dead blocks into the effective use
177 * of the Phi-input if the Phi is not in a bad block.
179 pred = get_nodes_block(n);
180 if (irn_not_visited(pred))
181 waitq_put(worklist, pred);
183 for (i = irn_arity - 1; i >= 0; --i) {
184 ir_node *pred = get_irn_n(n, i);
186 if (irn_not_visited(pred)) {
187 if (! in_dead_block &&
188 get_irn_pinned(pred) == op_pin_state_floats &&
189 is_Block_unreachable(get_nodes_block(pred))) {
190 set_nodes_block(pred, get_Block_cfgpred_block(curr_block, i));
192 waitq_put(worklist, pred);
197 ir_node *curr_block = get_nodes_block(n);
198 int in_dead_block = is_Block_unreachable(curr_block);
201 * All other nodes: move nodes from dead blocks into the same block.
203 pred = get_nodes_block(n);
204 if (irn_not_visited(pred))
205 waitq_put(worklist, pred);
207 for (i = irn_arity - 1; i >= 0; --i) {
208 ir_node *pred = get_irn_n(n, i);
210 if (irn_not_visited(pred)) {
211 if (! in_dead_block &&
212 get_irn_pinned(pred) == op_pin_state_floats &&
213 is_Block_unreachable(get_nodes_block(pred))) {
214 set_nodes_block(pred, curr_block);
216 waitq_put(worklist, pred);
223 * Floating nodes form subgraphs that begin at nodes as Const, Load,
224 * Start, Call and that end at op_pin_state_pinned nodes as Store, Call. Place_early
225 * places all floating nodes reachable from its argument through floating
226 * nodes and adds all beginnings at op_pin_state_pinned nodes to the worklist.
228 * @param worklist a worklist, used for the algorithm, empty on in/output
230 static void place_early(waitq *worklist) {
232 inc_irg_visited(current_ir_graph);
234 /* this inits the worklist */
235 place_floats_early(get_irg_end(current_ir_graph), worklist);
237 /* Work the content of the worklist. */
238 while (!waitq_empty(worklist)) {
239 ir_node *n = waitq_get(worklist);
240 if (irn_not_visited(n))
241 place_floats_early(n, worklist);
243 set_irg_pinned(current_ir_graph, op_pin_state_pinned);
247 * Compute the deepest common dominator tree ancestor of block and dca.
249 * @param dca the deepest common dominator tree ancestor so far,
251 * @param block a block
253 * @return the deepest common dominator tree ancestor of block and dca
255 static ir_node *calc_dom_dca(ir_node *dca, ir_node *block) {
258 /* we do not want to place nodes in dead blocks */
259 if (is_Block_dead(block))
262 /* We found a first legal placement. */
263 if (!dca) return block;
265 /* Find a placement that is dominates both, dca and block. */
266 while (get_Block_dom_depth(block) > get_Block_dom_depth(dca))
267 block = get_Block_idom(block);
269 while (get_Block_dom_depth(dca) > get_Block_dom_depth(block)) {
270 dca = get_Block_idom(dca);
273 while (block != dca) {
274 block = get_Block_idom(block); dca = get_Block_idom(dca);
280 * Deepest common dominance ancestor of DCA and CONSUMER of PRODUCER.
281 * I.e., DCA is the block where we might place PRODUCER.
282 * A data flow edge points from producer to consumer.
284 static ir_node *consumer_dom_dca(ir_node *dca, ir_node *consumer, ir_node *producer)
286 /* Compute the last block into which we can place a node so that it is
288 if (is_Phi(consumer)) {
289 /* our consumer is a Phi-node, the effective use is in all those
290 blocks through which the Phi-node reaches producer */
291 ir_node *phi_block = get_nodes_block(consumer);
292 int arity = get_irn_arity(consumer);
295 for (i = 0; i < arity; i++) {
296 if (get_Phi_pred(consumer, i) == producer) {
297 ir_node *new_block = get_Block_cfgpred_block(phi_block, i);
299 if (!is_Block_unreachable(new_block))
300 dca = calc_dom_dca(dca, new_block);
304 dca = calc_dom_dca(dca, get_nodes_block(consumer));
309 /* FIXME: the name clashes here with the function from ana/field_temperature.c
311 static INLINE int get_irn_loop_depth(ir_node *n) {
312 return get_loop_depth(get_irn_loop(n));
316 * Move n to a block with less loop depth than it's current block. The
317 * new block must be dominated by early.
319 * @param n the node that should be moved
320 * @param early the earliest block we can n move to
322 static void move_out_of_loops(ir_node *n, ir_node *early) {
327 /* Find the region deepest in the dominator tree dominating
328 dca with the least loop nesting depth, but still dominated
329 by our early placement. */
330 dca = get_nodes_block(n);
333 while (dca != early) {
334 dca = get_Block_idom(dca);
335 if (!dca || is_Bad(dca)) break; /* may be Bad if not reachable from Start */
336 if (get_irn_loop_depth(dca) < get_irn_loop_depth(best)) {
340 if (best != get_nodes_block(n))
341 set_nodes_block(n, best);
345 * Calculate the deepest common ancestor in the dominator tree of all nodes'
346 * blocks depending on node; our final placement has to dominate DCA.
348 * @param node the definition node
349 * @param dca the deepest common ancestor block so far, initially
352 * @return the deepest common dominator ancestor of all blocks of node's users
354 static ir_node *get_deepest_common_dom_ancestor(ir_node *node, ir_node *dca) {
357 for (i = get_irn_n_outs(node) - 1; i >= 0; --i) {
358 ir_node *succ = get_irn_out(node, i);
362 * This consumer is the End node, a keep alive edge.
363 * This is not a real consumer, so we ignore it
369 /* Proj nodes are in the same block as node, so
370 * the users of Proj are our users. */
371 dca = get_deepest_common_dom_ancestor(succ, dca);
373 /* ignore if succ is in dead code */
374 ir_node *succ_blk = get_nodes_block(succ);
375 if (is_Block_unreachable(succ_blk))
377 dca = consumer_dom_dca(dca, succ, node);
384 * Put all the Proj nodes of a node into a given block.
386 * @param node the mode_T node
387 * @param block the block to put the Proj nodes to
389 static void set_projs_block(ir_node *node, ir_node *block) {
392 for (i = get_irn_n_outs(node) - 1; i >= 0; --i) {
393 ir_node *succ = get_irn_out(node, i);
395 assert(is_Proj(succ));
397 if (get_irn_mode(succ) == mode_T) {
398 set_projs_block(succ, block);
400 set_nodes_block(succ, block);
405 * Find the latest legal block for N and place N into the
406 * `optimal' Block between the latest and earliest legal block.
407 * The `optimal' block is the dominance-deepest block of those
408 * with the least loop-nesting-depth. This places N out of as many
409 * loops as possible and then makes it as control dependent as
412 * @param n the node to be placed
413 * @param worklist a worklist, all successors of non-floating nodes are
416 static void place_floats_late(ir_node *n, pdeq *worklist) {
420 assert(irn_not_visited(n)); /* no multiple placement */
424 /* no need to place block nodes, control nodes are already placed. */
427 (get_irn_mode(n) != mode_X)) {
428 /* Remember the early_blk placement of this block to move it
429 out of loop no further than the early_blk placement. */
430 early_blk = get_nodes_block(n);
433 * BEWARE: Here we also get code, that is live, but
434 * was in a dead block. If the node is life, but because
435 * of CSE in a dead block, we still might need it.
438 /* Assure that our users are all placed, except the Phi-nodes.
439 --- Each data flow cycle contains at least one Phi-node. We
440 have to break the `user has to be placed before the
441 producer' dependence cycle and the Phi-nodes are the
442 place to do so, because we need to base our placement on the
443 final region of our users, which is OK with Phi-nodes, as they
444 are op_pin_state_pinned, and they never have to be placed after a
445 producer of one of their inputs in the same block anyway. */
446 for (i = get_irn_n_outs(n) - 1; i >= 0; --i) {
447 ir_node *succ = get_irn_out(n, i);
448 if (irn_not_visited(succ) && !is_Phi(succ))
449 place_floats_late(succ, worklist);
452 if (! is_Block_dead(early_blk)) {
453 /* do only move things that where not dead */
454 ir_op *op = get_irn_op(n);
456 /* We have to determine the final block of this node... except for
457 constants and Projs */
458 if ((get_irn_pinned(n) == op_pin_state_floats) &&
460 (op != op_SymConst) &&
463 /* deepest common ancestor in the dominator tree of all nodes'
464 blocks depending on us; our final placement has to dominate
466 ir_node *dca = get_deepest_common_dom_ancestor(n, NULL);
468 set_nodes_block(n, dca);
469 move_out_of_loops(n, early_blk);
470 if (get_irn_mode(n) == mode_T) {
471 set_projs_block(n, get_nodes_block(n));
478 /* Add successors of all non-floating nodes on list. (Those of floating
479 nodes are placed already and therefore are marked.) */
480 n_outs = get_irn_n_outs(n);
481 for (i = 0; i < n_outs; i++) {
482 ir_node *succ = get_irn_out(n, i);
483 if (irn_not_visited(succ)) {
484 pdeq_putr(worklist, succ);
490 * Place floating nodes on the given worklist as late as possible using
491 * the dominance tree.
493 * @param worklist the worklist containing the nodes to place
495 static void place_late(waitq *worklist) {
497 inc_irg_visited(current_ir_graph);
499 /* This fills the worklist initially. */
500 place_floats_late(get_irg_start_block(current_ir_graph), worklist);
502 /* And now empty the worklist again... */
503 while (!waitq_empty(worklist)) {
504 ir_node *n = waitq_get(worklist);
505 if (irn_not_visited(n))
506 place_floats_late(n, worklist);
510 /* Code Placement. */
511 void place_code(ir_graph *irg) {
513 ir_graph *rem = current_ir_graph;
515 current_ir_graph = irg;
517 /* Handle graph state */
518 assert(get_irg_phase_state(irg) != phase_building);
521 if (1 || get_irg_loopinfo_state(irg) != loopinfo_consistent) {
522 free_loop_information(irg);
523 construct_cf_backedges(irg);
526 /* Place all floating nodes as early as possible. This guarantees
527 a legal code placement. */
528 worklist = new_waitq();
529 place_early(worklist);
531 /* Note: place_early changes only blocks, no data edges. So, the
532 * data out edges are still valid, no need to recalculate them here. */
534 /* Now move the nodes down in the dominator tree. This reduces the
535 unnecessary executions of the node. */
536 place_late(worklist);
538 set_irg_outs_inconsistent(irg);
539 set_irg_loopinfo_inconsistent(irg);
541 current_ir_graph = rem;