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 Simple copy minimization heuristics.
23 * @author Christian Wuerdig
27 * This is the C implementation of the mst algorithm
28 * originally written in Java by Sebastian Hack.
29 * (also known as "heur3" :)
30 * Performs simple copy minimization.
34 #endif /* HAVE_CONFIG_H */
36 #define DISABLE_STATEV
43 #include "raw_bitset.h"
44 #include "irphase_t.h"
60 #include "becopyopt_t.h"
64 #define COL_COST_INFEASIBLE DBL_MAX
65 #define AFF_NEIGHBOUR_FIX_BENEFIT 128.0
66 #define NEIGHBOUR_CONSTR_COSTS 64.0
71 #define DBG_AFF_CHUNK(env, level, chunk) do { if (firm_dbg_get_mask(dbg) & (level)) dbg_aff_chunk((env), (chunk)); } while(0)
72 #define DBG_COL_COST(env, level, cost) do { if (firm_dbg_get_mask(dbg) & (level)) dbg_col_cost((env), (cost)); } while(0)
74 static firm_dbg_module_t *dbg = NULL;
78 #define DBG_AFF_CHUNK(env, level, chunk)
79 #define DBG_COL_COST(env, level, cost)
84 #define REAL(C) (C ## f)
86 static unsigned last_chunk_id = 0;
87 static int recolor_limit = 7;
88 static real_t dislike_influence = REAL(0.1);
90 typedef struct _col_cost_t {
98 typedef struct _aff_chunk_t {
99 const ir_node **n; /**< An ARR_F containing all nodes of the chunk. */
100 const ir_node **interfere; /**< An ARR_F containing all inference. */
101 int weight; /**< Weight of this chunk */
102 unsigned weight_consistent : 1; /**< Set if the weight is consistent. */
103 unsigned deleted : 1; /**< For debugging: Set if the was deleted. */
104 unsigned id; /**< An id of this chunk. */
106 col_cost_t color_affinity[1];
112 typedef struct _aff_edge_t {
113 const ir_node *src; /**< Source node. */
114 const ir_node *tgt; /**< Target node. */
115 int weight; /**< The weight of this edge. */
118 /* main coalescing environment */
119 typedef struct _co_mst_env_t {
120 int n_regs; /**< number of regs in class */
121 int k; /**< number of non-ignore registers in class */
122 bitset_t *ignore_regs; /**< set containing all global ignore registers */
123 ir_phase ph; /**< phase object holding data for nodes */
124 pqueue_t *chunks; /**< priority queue for chunks */
125 pset *chunkset; /**< set holding all chunks */
126 be_ifg_t *ifg; /**< the interference graph */
127 copy_opt_t *co; /**< the copy opt object */
128 unsigned chunk_visited;
129 col_cost_t **single_cols;
132 /* stores coalescing related information for a node */
133 typedef struct _co_mst_irn_t {
134 const ir_node *irn; /**< the irn this information belongs to */
135 aff_chunk_t *chunk; /**< the chunk this irn belongs to */
136 bitset_t *adm_colors; /**< set of admissible colors for this irn */
137 ir_node **int_neighs; /**< array of all interfering neighbours (cached for speed reasons) */
138 int n_neighs; /**< length of the interfering neighbours array. */
139 int int_aff_neigh; /**< number of interfering affinity neighbours */
140 int col; /**< color currently assigned */
141 int init_col; /**< the initial color */
142 int tmp_col; /**< a temporary assigned color */
143 unsigned fixed : 1; /**< the color is fixed */
144 struct list_head list; /**< Queue for coloring undo. */
145 real_t constr_factor;
148 #define get_co_mst_irn(mst_env, irn) (phase_get_or_set_irn_data(&(mst_env)->ph, (irn)))
150 typedef int decide_func_t(const co_mst_irn_t *node, int col);
155 * Write a chunk to stderr for debugging.
157 static void dbg_aff_chunk(const co_mst_env_t *env, const aff_chunk_t *c) {
160 if (c->weight_consistent)
161 ir_fprintf(stderr, " $%d ", c->weight);
162 ir_fprintf(stderr, "{");
163 for (i = 0, l = ARR_LEN(c->n); i < l; ++i) {
164 const ir_node *n = c->n[i];
165 ir_fprintf(stderr, " %+F,", n);
167 ir_fprintf(stderr, "}");
171 * Dump all admissible colors to stderr.
173 static void dbg_admissible_colors(const co_mst_env_t *env, const co_mst_irn_t *node) {
177 if (bitset_popcnt(node->adm_colors) < 1)
178 fprintf(stderr, "no admissible colors?!?");
180 bitset_foreach(node->adm_colors, idx) {
181 fprintf(stderr, " %d", idx);
187 * Dump color-cost pairs to stderr.
189 static void dbg_col_cost(const co_mst_env_t *env, const col_cost_t *cost) {
191 for (i = 0; i < env->n_regs; ++i)
192 fprintf(stderr, " (%d, %.4f)", cost[i].col, cost[i].cost);
195 #endif /* DEBUG_libfirm */
197 static INLINE int get_mst_irn_col(const co_mst_irn_t *node) {
198 return node->tmp_col >= 0 ? node->tmp_col : node->col;
202 * @return 1 if node @p node has color @p col, 0 otherwise.
204 static int decider_has_color(const co_mst_irn_t *node, int col) {
205 return get_mst_irn_col(node) == col;
209 * @return 1 if node @p node has not color @p col, 0 otherwise.
211 static int decider_hasnot_color(const co_mst_irn_t *node, int col) {
212 return get_mst_irn_col(node) != col;
216 * Always returns true.
218 static int decider_always_yes(const co_mst_irn_t *node, int col) {
224 /** compares two affinity edges by its weight */
225 static int cmp_aff_edge(const void *a, const void *b) {
226 const aff_edge_t *e1 = a;
227 const aff_edge_t *e2 = b;
229 if (e2->weight == e1->weight) {
230 if (e2->src->node_idx == e1->src->node_idx)
231 return QSORT_CMP(e2->tgt->node_idx, e1->tgt->node_idx);
233 return QSORT_CMP(e2->src->node_idx, e1->src->node_idx);
235 /* sort in descending order */
236 return QSORT_CMP(e2->weight, e1->weight);
239 /** compares to color-cost pairs */
240 static __attribute__((unused)) int cmp_col_cost_lt(const void *a, const void *b) {
241 const col_cost_t *c1 = a;
242 const col_cost_t *c2 = b;
243 real_t diff = c1->cost - c2->cost;
244 return (diff > 0) - (diff < 0);
247 static int cmp_col_cost_gt(const void *a, const void *b) {
248 const col_cost_t *c1 = a;
249 const col_cost_t *c2 = b;
250 real_t diff = c2->cost - c1->cost;
251 return (diff > 0) - (diff < 0);
255 * Creates a new affinity chunk
257 static INLINE aff_chunk_t *new_aff_chunk(co_mst_env_t *env) {
258 aff_chunk_t *c = XMALLOCF(aff_chunk_t, color_affinity, env->n_regs);
259 c->n = NEW_ARR_F(const ir_node *, 0);
260 c->interfere = NEW_ARR_F(const ir_node *, 0);
262 c->weight_consistent = 0;
264 c->id = ++last_chunk_id;
266 pset_insert(env->chunkset, c, c->id);
271 * Frees all memory allocated by an affinity chunk.
273 static INLINE void delete_aff_chunk(co_mst_env_t *env, aff_chunk_t *c) {
274 pset_remove(env->chunkset, c, c->id);
275 DEL_ARR_F(c->interfere);
282 * binary search of sorted nodes.
284 * @return the position where n is found in the array arr or ~pos
285 * if the nodes is not here.
287 static INLINE int nodes_bsearch(const ir_node **arr, const ir_node *n) {
288 int hi = ARR_LEN(arr);
292 int md = lo + ((hi - lo) >> 1);
305 /** Check if a node n can be found inside arr. */
306 static int node_contains(const ir_node **arr, const ir_node *n) {
307 int i = nodes_bsearch(arr, n);
312 * Insert a node into the sorted nodes list.
314 * @return 1 if the node was inserted, 0 else
316 static int nodes_insert(const ir_node ***arr, const ir_node *irn) {
317 int idx = nodes_bsearch(*arr, irn);
320 int i, n = ARR_LEN(*arr);
323 ARR_APP1(const ir_node *, *arr, irn);
328 for (i = n - 1; i >= idx; --i)
337 * Adds a node to an affinity chunk
339 static INLINE void aff_chunk_add_node(aff_chunk_t *c, co_mst_irn_t *node) {
342 if (! nodes_insert(&c->n, node->irn))
345 c->weight_consistent = 0;
348 for (i = node->n_neighs - 1; i >= 0; --i) {
349 ir_node *neigh = node->int_neighs[i];
350 nodes_insert(&c->interfere, neigh);
355 * In case there is no phase information for irn, initialize it.
357 static void *co_mst_irn_init(ir_phase *ph, const ir_node *irn, void *old) {
358 co_mst_irn_t *res = old ? old : phase_alloc(ph, sizeof(res[0]));
359 co_mst_env_t *env = ph->priv;
362 const arch_register_req_t *req;
363 void *nodes_it = be_ifg_nodes_iter_alloca(env->ifg);
371 res->int_neighs = NULL;
372 res->int_aff_neigh = 0;
373 res->col = arch_register_get_index(arch_get_irn_register(irn));
374 res->init_col = res->col;
375 INIT_LIST_HEAD(&res->list);
377 DB((dbg, LEVEL_4, "Creating phase info for %+F\n", irn));
379 /* set admissible registers */
380 res->adm_colors = bitset_obstack_alloc(phase_obst(ph), env->n_regs);
382 /* Exclude colors not assignable to the irn */
383 req = arch_get_register_req(irn, -1);
384 if (arch_register_req_is(req, limited))
385 rbitset_copy_to_bitset(req->limited, res->adm_colors);
387 bitset_set_all(res->adm_colors);
389 /* exclude global ignore registers as well */
390 bitset_andnot(res->adm_colors, env->ignore_regs);
392 /* compute the constraint factor */
393 res->constr_factor = (real_t) (1 + env->n_regs - bitset_popcnt(res->adm_colors)) / env->n_regs;
395 /* set the number of interfering affinity neighbours to -1, they are calculated later */
396 res->int_aff_neigh = -1;
398 /* build list of interfering neighbours */
400 be_ifg_foreach_neighbour(env->ifg, nodes_it, irn, neigh) {
401 if (!arch_irn_is(neigh, ignore)) {
402 obstack_ptr_grow(phase_obst(ph), neigh);
406 res->int_neighs = obstack_finish(phase_obst(ph));
413 * Check if affinity chunk @p chunk interferes with node @p irn.
415 static INLINE int aff_chunk_interferes(const aff_chunk_t *chunk, const ir_node *irn) {
416 return node_contains(chunk->interfere, irn);
420 * Check if there are interference edges from c1 to c2.
422 * @param c2 Another chunk
423 * @return 1 if there are interferences between nodes of c1 and c2, 0 otherwise.
425 static INLINE int aff_chunks_interfere(const aff_chunk_t *c1, const aff_chunk_t *c2) {
431 /* check if there is a node in c2 having an interfering neighbor in c1 */
432 for (i = ARR_LEN(c2->n) - 1; i >= 0; --i) {
433 const ir_node *irn = c2->n[i];
435 if (node_contains(c1->interfere, irn))
442 * Returns the affinity chunk of @p irn or creates a new
443 * one with @p irn as element if there is none assigned.
445 static INLINE aff_chunk_t *get_aff_chunk(co_mst_env_t *env, const ir_node *irn) {
446 co_mst_irn_t *node = get_co_mst_irn(env, irn);
451 * Let chunk(src) absorb the nodes of chunk(tgt) (only possible when there
452 * are no interference edges from chunk(src) to chunk(tgt)).
453 * @return 1 if successful, 0 if not possible
455 static int aff_chunk_absorb(co_mst_env_t *env, const ir_node *src, const ir_node *tgt) {
456 aff_chunk_t *c1 = get_aff_chunk(env, src);
457 aff_chunk_t *c2 = get_aff_chunk(env, tgt);
460 DB((dbg, LEVEL_4, "Attempt to let c1 (id %u): ", c1 ? c1->id : 0));
462 DBG_AFF_CHUNK(env, LEVEL_4, c1);
464 DB((dbg, LEVEL_4, "{%+F}", src));
466 DB((dbg, LEVEL_4, "\n\tabsorb c2 (id %u): ", c2 ? c2->id : 0));
468 DBG_AFF_CHUNK(env, LEVEL_4, c2);
470 DB((dbg, LEVEL_4, "{%+F}", tgt));
472 DB((dbg, LEVEL_4, "\n"));
477 /* no chunk exists */
478 co_mst_irn_t *mirn = get_co_mst_irn(env, src);
481 for (i = mirn->n_neighs - 1; i >= 0; --i) {
482 if (mirn->int_neighs[i] == tgt)
486 /* create one containing both nodes */
487 c1 = new_aff_chunk(env);
488 aff_chunk_add_node(c1, get_co_mst_irn(env, src));
489 aff_chunk_add_node(c1, get_co_mst_irn(env, tgt));
493 /* c2 already exists */
494 if (! aff_chunk_interferes(c2, src)) {
495 aff_chunk_add_node(c2, get_co_mst_irn(env, src));
499 } else if (c2 == NULL) {
500 /* c1 already exists */
501 if (! aff_chunk_interferes(c1, tgt)) {
502 aff_chunk_add_node(c1, get_co_mst_irn(env, tgt));
505 } else if (c1 != c2 && ! aff_chunks_interfere(c1, c2)) {
508 for (idx = 0, len = ARR_LEN(c2->n); idx < len; ++idx)
509 aff_chunk_add_node(c1, get_co_mst_irn(env, c2->n[idx]));
511 for (idx = 0, len = ARR_LEN(c2->interfere); idx < len; ++idx) {
512 const ir_node *irn = c2->interfere[idx];
513 nodes_insert(&c1->interfere, irn);
516 c1->weight_consistent = 0;
518 delete_aff_chunk(env, c2);
521 DB((dbg, LEVEL_4, " ... c1 interferes with c2, skipped\n"));
525 DB((dbg, LEVEL_4, " ... absorbed\n"));
530 * Assures that the weight of the given chunk is consistent.
532 static void aff_chunk_assure_weight(co_mst_env_t *env, aff_chunk_t *c) {
533 if (! c->weight_consistent) {
537 for (i = 0; i < env->n_regs; ++i) {
538 c->color_affinity[i].col = i;
539 c->color_affinity[i].cost = REAL(0.0);
542 for (idx = 0, len = ARR_LEN(c->n); idx < len; ++idx) {
543 const ir_node *n = c->n[idx];
544 const affinity_node_t *an = get_affinity_info(env->co, n);
545 co_mst_irn_t *node = get_co_mst_irn(env, n);
548 if (node->constr_factor > REAL(0.0)) {
550 bitset_foreach (node->adm_colors, col)
551 c->color_affinity[col].cost += node->constr_factor;
556 co_gs_foreach_neighb(an, neigh) {
557 const ir_node *m = neigh->irn;
559 /* skip ignore nodes */
560 if (arch_irn_is(m, ignore))
563 w += node_contains(c->n, m) ? neigh->costs : 0;
568 for (i = 0; i < env->n_regs; ++i)
569 c->color_affinity[i].cost *= (REAL(1.0) / ARR_LEN(c->n));
572 // c->weight = bitset_popcnt(c->nodes);
573 c->weight_consistent = 1;
578 * Count the number of interfering affinity neighbours
580 static int count_interfering_aff_neighs(co_mst_env_t *env, const affinity_node_t *an) {
581 const neighb_t *neigh;
582 const ir_node *irn = an->irn;
583 const co_mst_irn_t *node = get_co_mst_irn(env, irn);
586 co_gs_foreach_neighb(an, neigh) {
587 const ir_node *n = neigh->irn;
590 /* skip ignore nodes */
591 if (arch_irn_is(n, ignore))
594 /* check if the affinity neighbour interfere */
595 for (i = 0; i < node->n_neighs; ++i) {
596 if (node->int_neighs[i] == n) {
607 * Build chunks of nodes connected by affinity edges.
608 * We start at the heaviest affinity edge.
609 * The chunks of the two edge-defining nodes will be
610 * merged if there are no interference edges from one
611 * chunk to the other.
613 static void build_affinity_chunks(co_mst_env_t *env) {
614 void *nodes_it = be_ifg_nodes_iter_alloca(env->ifg);
615 aff_edge_t *edges = NEW_ARR_F(aff_edge_t, 0);
618 aff_chunk_t *curr_chunk;
620 /* at first we create the affinity edge objects */
621 be_ifg_foreach_node(env->ifg, nodes_it, n) {
622 int n_idx = get_irn_idx(n);
626 /* skip ignore nodes */
627 if (arch_irn_is(n, ignore))
630 n1 = get_co_mst_irn(env, n);
631 an = get_affinity_info(env->co, n);
636 if (n1->int_aff_neigh < 0)
637 n1->int_aff_neigh = count_interfering_aff_neighs(env, an);
639 /* build the affinity edges */
640 co_gs_foreach_neighb(an, neigh) {
641 const ir_node *m = neigh->irn;
642 int m_idx = get_irn_idx(m);
644 /* record the edge in only one direction */
649 /* skip ignore nodes */
650 if (arch_irn_is(m, ignore))
656 n2 = get_co_mst_irn(env, m);
657 if (n2->int_aff_neigh < 0) {
658 affinity_node_t *am = get_affinity_info(env->co, m);
659 n2->int_aff_neigh = count_interfering_aff_neighs(env, am);
662 * these weights are pure hackery ;-).
663 * It's not chriswue's fault but mine.
665 edge.weight = neigh->costs;
666 ARR_APP1(aff_edge_t, edges, edge);
672 /* now: sort edges and build the affinity chunks */
673 len = ARR_LEN(edges);
674 qsort(edges, len, sizeof(edges[0]), cmp_aff_edge);
675 for (i = 0; i < len; ++i) {
676 DBG((dbg, LEVEL_1, "edge (%u,%u) %f\n", edges[i].src->node_idx, edges[i].tgt->node_idx, edges[i].weight));
678 (void)aff_chunk_absorb(env, edges[i].src, edges[i].tgt);
681 /* now insert all chunks into a priority queue */
682 foreach_pset(env->chunkset, curr_chunk) {
683 aff_chunk_assure_weight(env, curr_chunk);
685 DBG((dbg, LEVEL_1, "entry #%u", curr_chunk->id));
686 DBG_AFF_CHUNK(env, LEVEL_1, curr_chunk);
687 DBG((dbg, LEVEL_1, "\n"));
689 pqueue_put(env->chunks, curr_chunk, curr_chunk->weight);
692 foreach_phase_irn(&env->ph, n) {
693 co_mst_irn_t *mirn = get_co_mst_irn(env, n);
695 if (mirn->chunk == NULL) {
696 /* no chunk is allocated so far, do it now */
697 aff_chunk_t *curr_chunk = new_aff_chunk(env);
698 aff_chunk_add_node(curr_chunk, mirn);
700 aff_chunk_assure_weight(env, curr_chunk);
702 DBG((dbg, LEVEL_1, "entry #%u", curr_chunk->id));
703 DBG_AFF_CHUNK(env, LEVEL_1, curr_chunk);
704 DBG((dbg, LEVEL_1, "\n"));
706 pqueue_put(env->chunks, curr_chunk, curr_chunk->weight);
713 static __attribute__((unused)) void chunk_order_nodes(co_mst_env_t *env, aff_chunk_t *chunk)
715 pqueue_t *grow = new_pqueue();
716 const ir_node *max_node = NULL;
720 for (i = ARR_LEN(chunk->n) - 1; i >= 0; i--) {
721 const ir_node *irn = chunk->n[i];
722 affinity_node_t *an = get_affinity_info(env->co, irn);
726 if (arch_irn_is(irn, ignore))
730 co_gs_foreach_neighb(an, neigh)
733 if (w > max_weight) {
741 bitset_t *visited = bitset_irg_malloc(env->co->irg);
743 for (i = ARR_LEN(chunk->n) - 1; i >= 0; --i)
744 bitset_add_irn(visited, chunk->n[i]);
746 pqueue_put(grow, (void *) max_node, max_weight);
747 bitset_remv_irn(visited, max_node);
749 while (!pqueue_empty(grow)) {
750 ir_node *irn = pqueue_pop_front(grow);
751 affinity_node_t *an = get_affinity_info(env->co, irn);
754 if (arch_irn_is(irn, ignore))
757 assert(i <= ARR_LEN(chunk->n));
762 /* build the affinity edges */
763 co_gs_foreach_neighb(an, neigh) {
764 co_mst_irn_t *node = get_co_mst_irn(env, neigh->irn);
766 if (bitset_contains_irn(visited, node->irn)) {
767 pqueue_put(grow, (void *) neigh->irn, neigh->costs);
768 bitset_remv_irn(visited, node->irn);
774 bitset_free(visited);
779 * Greedy collect affinity neighbours into thew new chunk @p chunk starting at node @p node.
781 static void expand_chunk_from(co_mst_env_t *env, co_mst_irn_t *node, bitset_t *visited,
782 aff_chunk_t *chunk, aff_chunk_t *orig_chunk, decide_func_t *decider, int col)
784 waitq *nodes = new_waitq();
786 DBG((dbg, LEVEL_1, "\n\tExpanding new chunk (#%u) from %+F, color %d:", chunk->id, node->irn, col));
788 /* init queue and chunk */
789 waitq_put(nodes, node);
790 bitset_set(visited, get_irn_idx(node->irn));
791 aff_chunk_add_node(chunk, node);
792 DB((dbg, LEVEL_1, " %+F", node->irn));
794 /* as long as there are nodes in the queue */
795 while (! waitq_empty(nodes)) {
796 co_mst_irn_t *n = waitq_get(nodes);
797 affinity_node_t *an = get_affinity_info(env->co, n->irn);
799 /* check all affinity neighbors */
802 co_gs_foreach_neighb(an, neigh) {
803 const ir_node *m = neigh->irn;
804 int m_idx = get_irn_idx(m);
807 /* skip ignore nodes */
808 if (arch_irn_is(m, ignore))
811 n2 = get_co_mst_irn(env, m);
813 if (! bitset_is_set(visited, m_idx) &&
816 ! aff_chunk_interferes(chunk, m) &&
817 node_contains(orig_chunk->n, m))
820 following conditions are met:
821 - neighbour is not visited
822 - neighbour likes the color
823 - neighbour has not yet a fixed color
824 - the new chunk doesn't interfere with the neighbour
825 - neighbour belongs or belonged once to the original chunk
827 bitset_set(visited, m_idx);
828 aff_chunk_add_node(chunk, n2);
829 DB((dbg, LEVEL_1, " %+F", n2->irn));
830 /* enqueue for further search */
831 waitq_put(nodes, n2);
837 DB((dbg, LEVEL_1, "\n"));
843 * Fragment the given chunk into chunks having given color and not having given color.
845 static aff_chunk_t *fragment_chunk(co_mst_env_t *env, int col, aff_chunk_t *c, waitq *tmp) {
846 bitset_t *visited = bitset_irg_malloc(env->co->irg);
848 aff_chunk_t *best = NULL;
850 for (idx = 0, len = ARR_LEN(c->n); idx < len; ++idx) {
853 aff_chunk_t *tmp_chunk;
854 decide_func_t *decider;
858 if (bitset_is_set(visited, get_irn_idx(irn)))
861 node = get_co_mst_irn(env, irn);
863 if (get_mst_irn_col(node) == col) {
864 decider = decider_has_color;
866 DBG((dbg, LEVEL_4, "\tcolor %d wanted\n", col));
869 decider = decider_hasnot_color;
871 DBG((dbg, LEVEL_4, "\tcolor %d forbidden\n", col));
874 /* create a new chunk starting at current node */
875 tmp_chunk = new_aff_chunk(env);
876 waitq_put(tmp, tmp_chunk);
877 expand_chunk_from(env, node, visited, tmp_chunk, c, decider, col);
878 assert(ARR_LEN(tmp_chunk->n) > 0 && "No nodes added to chunk");
880 /* remember the local best */
881 aff_chunk_assure_weight(env, tmp_chunk);
882 if (check_for_best && (! best || best->weight < tmp_chunk->weight))
886 assert(best && "No chunk found?");
887 bitset_free(visited);
892 * Resets the temporary fixed color of all nodes within wait queue @p nodes.
893 * ATTENTION: the queue is empty after calling this function!
895 static INLINE void reject_coloring(struct list_head *nodes) {
896 co_mst_irn_t *n, *temp;
897 DB((dbg, LEVEL_4, "\treject coloring for"));
898 list_for_each_entry_safe(co_mst_irn_t, n, temp, nodes, list) {
899 DB((dbg, LEVEL_4, " %+F", n->irn));
900 assert(n->tmp_col >= 0);
902 list_del_init(&n->list);
904 DB((dbg, LEVEL_4, "\n"));
907 static INLINE void materialize_coloring(struct list_head *nodes) {
908 co_mst_irn_t *n, *temp;
909 list_for_each_entry_safe(co_mst_irn_t, n, temp, nodes, list) {
910 assert(n->tmp_col >= 0);
913 list_del_init(&n->list);
917 static INLINE void set_temp_color(co_mst_irn_t *node, int col, struct list_head *changed)
920 assert(!node->fixed);
921 assert(node->tmp_col < 0);
922 assert(node->list.next == &node->list && node->list.prev == &node->list);
923 assert(bitset_is_set(node->adm_colors, col));
925 list_add_tail(&node->list, changed);
929 static INLINE int is_loose(co_mst_irn_t *node)
931 return !node->fixed && node->tmp_col < 0;
935 * Determines the costs for each color if it would be assigned to node @p node.
937 static void determine_color_costs(co_mst_env_t *env, co_mst_irn_t *node, col_cost_t *costs) {
938 int *neigh_cols = alloca(env->n_regs * sizeof(*neigh_cols));
943 for (i = 0; i < env->n_regs; ++i) {
946 costs[i].cost = bitset_is_set(node->adm_colors, i) ? node->constr_factor : REAL(0.0);
949 for (i = 0; i < node->n_neighs; ++i) {
950 co_mst_irn_t *n = get_co_mst_irn(env, node->int_neighs[i]);
951 int col = get_mst_irn_col(n);
956 costs[col].cost = REAL(0.0);
960 coeff = REAL(1.0) / n_loose;
961 for (i = 0; i < env->n_regs; ++i)
962 costs[i].cost *= REAL(1.0) - coeff * neigh_cols[i];
966 /* need forward declaration due to recursive call */
967 static int recolor_nodes(co_mst_env_t *env, co_mst_irn_t *node, col_cost_t *costs, struct list_head *changed_ones, int depth, int *max_depth, int *trip);
970 * Tries to change node to a color but @p explude_col.
971 * @return 1 if succeeded, 0 otherwise.
973 static int change_node_color_excluded(co_mst_env_t *env, co_mst_irn_t *node, int exclude_col, struct list_head *changed, int depth, int *max_depth, int *trip) {
974 int col = get_mst_irn_col(node);
977 /* neighbours has already a different color -> good, temporary fix it */
978 if (col != exclude_col) {
980 set_temp_color(node, col, changed);
984 /* The node has the color it should not have _and_ has not been visited yet. */
985 if (is_loose(node)) {
986 col_cost_t *costs = alloca(env->n_regs * sizeof(costs[0]));
988 /* Get the costs for giving the node a specific color. */
989 determine_color_costs(env, node, costs);
991 /* Since the node must not have the not_col, set the costs for that color to "infinity" */
992 costs[exclude_col].cost = REAL(0.0);
994 /* sort the colors according costs, cheapest first. */
995 qsort(costs, env->n_regs, sizeof(costs[0]), cmp_col_cost_gt);
997 /* Try recoloring the node using the color list. */
998 res = recolor_nodes(env, node, costs, changed, depth + 1, max_depth, trip);
1005 * Tries to bring node @p node to cheapest color and color all interfering neighbours with other colors.
1006 * ATTENTION: Expect @p costs already sorted by increasing costs.
1007 * @return 1 if coloring could be applied, 0 otherwise.
1009 static int recolor_nodes(co_mst_env_t *env, co_mst_irn_t *node, col_cost_t *costs, struct list_head *changed, int depth, int *max_depth, int *trip) {
1011 struct list_head local_changed;
1014 if (depth > *max_depth)
1017 DBG((dbg, LEVEL_4, "\tRecoloring %+F with color-costs", node->irn));
1018 DBG_COL_COST(env, LEVEL_4, costs);
1019 DB((dbg, LEVEL_4, "\n"));
1021 if (depth >= recolor_limit) {
1022 DBG((dbg, LEVEL_4, "\tHit recolor limit\n"));
1026 for (i = 0; i < env->n_regs; ++i) {
1027 int tgt_col = costs[i].col;
1031 /* If the costs for that color (and all successive) are infinite, bail out we won't make it anyway. */
1032 if (costs[i].cost == REAL(0.0)) {
1033 DBG((dbg, LEVEL_4, "\tAll further colors forbidden\n"));
1037 /* Set the new color of the node and mark the node as temporarily fixed. */
1038 assert(node->tmp_col < 0 && "Node must not have been temporary fixed.");
1039 INIT_LIST_HEAD(&local_changed);
1040 set_temp_color(node, tgt_col, &local_changed);
1041 DBG((dbg, LEVEL_4, "\tTemporary setting %+F to color %d\n", node->irn, tgt_col));
1043 /* try to color all interfering neighbours with current color forbidden */
1044 for (j = 0; j < node->n_neighs; ++j) {
1048 neigh = node->int_neighs[j];
1050 /* skip ignore nodes */
1051 if (arch_irn_is(neigh, ignore))
1054 nn = get_co_mst_irn(env, neigh);
1055 DB((dbg, LEVEL_4, "\tHandling neighbour %+F, at position %d (fixed: %d, tmp_col: %d, col: %d)\n",
1056 neigh, j, nn->fixed, nn->tmp_col, nn->col));
1059 Try to change the color of the neighbor and record all nodes which
1060 get changed in the tmp list. Add this list to the "changed" list for
1061 that color. If we did not succeed to change the color of the neighbor,
1062 we bail out and try the next color.
1064 if (get_mst_irn_col(nn) == tgt_col) {
1065 /* try to color neighbour with tgt_col forbidden */
1066 neigh_ok = change_node_color_excluded(env, nn, tgt_col, &local_changed, depth + 1, max_depth, trip);
1074 We managed to assign the target color to all neighbors, so from the perspective
1075 of the current node, every thing was ok and we can return safely.
1078 /* append the local_changed ones to global ones */
1079 list_splice(&local_changed, changed);
1083 /* coloring of neighbours failed, so we try next color */
1084 reject_coloring(&local_changed);
1088 DBG((dbg, LEVEL_4, "\tAll colors failed\n"));
1093 * Tries to bring node @p node and all it's neighbours to color @p tgt_col.
1094 * @return 1 if color @p col could be applied, 0 otherwise
1096 static int change_node_color(co_mst_env_t *env, co_mst_irn_t *node, int tgt_col, struct list_head *changed) {
1097 int col = get_mst_irn_col(node);
1099 /* if node already has the target color -> good, temporary fix it */
1100 if (col == tgt_col) {
1101 DBG((dbg, LEVEL_4, "\t\tCNC: %+F has already color %d, fix temporary\n", node->irn, tgt_col));
1103 set_temp_color(node, tgt_col, changed);
1108 Node has not yet a fixed color and target color is admissible
1109 -> try to recolor node and it's affinity neighbours
1111 if (is_loose(node) && bitset_is_set(node->adm_colors, tgt_col)) {
1112 col_cost_t *costs = env->single_cols[tgt_col];
1113 int res, max_depth, trip;
1118 DBG((dbg, LEVEL_4, "\t\tCNC: Attempt to recolor %+F ===>>\n", node->irn));
1119 res = recolor_nodes(env, node, costs, changed, 0, &max_depth, &trip);
1120 DBG((dbg, LEVEL_4, "\t\tCNC: <<=== Recoloring of %+F %s\n", node->irn, res ? "succeeded" : "failed"));
1121 stat_ev_int("heur4_recolor_depth_max", max_depth);
1122 stat_ev_int("heur4_recolor_trip", trip);
1128 #ifdef DEBUG_libfirm
1129 if (firm_dbg_get_mask(dbg) & LEVEL_4) {
1130 if (!is_loose(node))
1131 DB((dbg, LEVEL_4, "\t\tCNC: %+F has already fixed color %d\n", node->irn, col));
1133 DB((dbg, LEVEL_4, "\t\tCNC: color %d not admissible for %+F (", tgt_col, node->irn));
1134 dbg_admissible_colors(env, node);
1135 DB((dbg, LEVEL_4, ")\n"));
1144 * Tries to color an affinity chunk (or at least a part of it).
1145 * Inserts uncolored parts of the chunk as a new chunk into the priority queue.
1147 static void color_aff_chunk(co_mst_env_t *env, aff_chunk_t *c) {
1148 aff_chunk_t *best_chunk = NULL;
1149 int n_nodes = ARR_LEN(c->n);
1150 int best_color = -1;
1151 int n_int_chunks = 0;
1152 waitq *tmp_chunks = new_waitq();
1153 waitq *best_starts = NULL;
1154 col_cost_t *order = alloca(env->n_regs * sizeof(order[0]));
1156 int idx, len, i, nidx, pos;
1157 struct list_head changed;
1159 DB((dbg, LEVEL_2, "fragmentizing chunk #%u", c->id));
1160 DBG_AFF_CHUNK(env, LEVEL_2, c);
1161 DB((dbg, LEVEL_2, "\n"));
1163 stat_ev_ctx_push_fmt("heur4_color_chunk", "%u", c->id);
1165 ++env->chunk_visited;
1167 /* compute color preference */
1168 memset(order, 0, env->n_regs * sizeof(order[0]));
1170 for (pos = 0, len = ARR_LEN(c->interfere); pos < len; ++pos) {
1171 const ir_node *n = c->interfere[pos];
1172 co_mst_irn_t *node = get_co_mst_irn(env, n);
1173 aff_chunk_t *chunk = node->chunk;
1175 if (is_loose(node) && chunk && chunk->visited < env->chunk_visited) {
1176 assert(!chunk->deleted);
1177 chunk->visited = env->chunk_visited;
1180 aff_chunk_assure_weight(env, chunk);
1181 for (i = 0; i < env->n_regs; ++i)
1182 order[i].cost += chunk->color_affinity[i].cost;
1186 for (i = 0; i < env->n_regs; ++i) {
1187 real_t dislike = n_int_chunks > 0 ? REAL(1.0) - order[i].cost / n_int_chunks : REAL(0.0);
1189 order[i].cost = (REAL(1.0) - dislike_influence) * c->color_affinity[i].cost + dislike_influence * dislike;
1192 qsort(order, env->n_regs, sizeof(order[0]), cmp_col_cost_gt);
1194 DBG_COL_COST(env, LEVEL_2, order);
1195 DB((dbg, LEVEL_2, "\n"));
1197 /* check which color is the "best" for the given chunk.
1198 * if we found a color which was ok for all nodes, we take it
1199 * and do not look further. (see did_all flag usage below.)
1200 * If we have many colors which fit all nodes it is hard to decide
1201 * which one to take anyway.
1202 * TODO Sebastian: Perhaps we should at all nodes and figure out
1203 * a suitable color using costs as done above (determine_color_costs).
1205 for (i = 0; i < env->k; ++i) {
1206 int col = order[i].col;
1207 waitq *good_starts = new_waitq();
1208 aff_chunk_t *local_best;
1211 /* skip ignore colors */
1212 if (bitset_is_set(env->ignore_regs, col))
1215 DB((dbg, LEVEL_2, "\ttrying color %d\n", col));
1219 /* try to bring all nodes of given chunk to the current color. */
1220 for (idx = 0, len = ARR_LEN(c->n); idx < len; ++idx) {
1221 const ir_node *irn = c->n[idx];
1222 co_mst_irn_t *node = get_co_mst_irn(env, irn);
1225 assert(! node->fixed && "Node must not have a fixed color.");
1226 DB((dbg, LEVEL_4, "\t\tBringing %+F from color %d to color %d ...\n", irn, node->col, col));
1229 The order of the colored nodes is important, so we record the successfully
1230 colored ones in the order they appeared.
1232 INIT_LIST_HEAD(&changed);
1234 good = change_node_color(env, node, col, &changed);
1235 stat_ev_tim_pop("heur4_recolor");
1237 waitq_put(good_starts, node);
1238 materialize_coloring(&changed);
1243 reject_coloring(&changed);
1245 n_succeeded += good;
1246 DB((dbg, LEVEL_4, "\t\t... %+F attempt from %d to %d %s\n", irn, node->col, col, good ? "succeeded" : "failed"));
1249 /* unfix all nodes */
1250 for (idx = 0, len = ARR_LEN(c->n); idx < len; ++idx) {
1251 co_mst_irn_t *node = get_co_mst_irn(env, c->n[idx]);
1255 /* try next color when failed */
1256 if (n_succeeded == 0)
1259 /* fragment the chunk according to the coloring */
1260 local_best = fragment_chunk(env, col, c, tmp_chunks);
1262 /* search the best of the good list
1263 and make it the new best if it is better than the current */
1265 aff_chunk_assure_weight(env, local_best);
1267 DB((dbg, LEVEL_3, "\t\tlocal best chunk (id %u) for color %d: ", local_best->id, col));
1268 DBG_AFF_CHUNK(env, LEVEL_3, local_best);
1270 if (! best_chunk || best_chunk->weight < local_best->weight) {
1271 best_chunk = local_best;
1274 del_waitq(best_starts);
1275 best_starts = good_starts;
1276 DB((dbg, LEVEL_3, "\n\t\t... setting global best chunk (id %u), color %d\n", best_chunk->id, best_color));
1278 DB((dbg, LEVEL_3, "\n\t\t... omitting, global best is better\n"));
1279 del_waitq(good_starts);
1283 del_waitq(good_starts);
1286 /* if all nodes were recolored, bail out */
1287 if (n_succeeded == n_nodes)
1291 stat_ev_int("heur4_colors_tried", i);
1293 /* free all intermediate created chunks except best one */
1294 while (! waitq_empty(tmp_chunks)) {
1295 aff_chunk_t *tmp = waitq_get(tmp_chunks);
1296 if (tmp != best_chunk)
1297 delete_aff_chunk(env, tmp);
1299 del_waitq(tmp_chunks);
1301 /* return if coloring failed */
1304 del_waitq(best_starts);
1308 DB((dbg, LEVEL_2, "\tbest chunk #%u ", best_chunk->id));
1309 DBG_AFF_CHUNK(env, LEVEL_2, best_chunk);
1310 DB((dbg, LEVEL_2, "using color %d\n", best_color));
1312 for (idx = 0, len = ARR_LEN(best_chunk->n); idx < len; ++idx) {
1313 const ir_node *irn = best_chunk->n[idx];
1314 co_mst_irn_t *node = get_co_mst_irn(env, irn);
1317 /* bring the node to the color. */
1318 DB((dbg, LEVEL_4, "\tManifesting color %d for %+F, chunk #%u\n", best_color, node->irn, best_chunk->id));
1319 INIT_LIST_HEAD(&changed);
1321 res = change_node_color(env, node, best_color, &changed);
1322 stat_ev_tim_pop("heur4_recolor");
1324 materialize_coloring(&changed);
1327 assert(list_empty(&changed));
1330 /* remove the nodes in best chunk from original chunk */
1331 len = ARR_LEN(best_chunk->n);
1332 for (idx = 0; idx < len; ++idx) {
1333 const ir_node *irn = best_chunk->n[idx];
1334 int pos = nodes_bsearch(c->n, irn);
1339 len = ARR_LEN(c->n);
1340 for (idx = nidx = 0; idx < len; ++idx) {
1341 const ir_node *irn = c->n[idx];
1347 ARR_SHRINKLEN(c->n, nidx);
1350 /* we have to get the nodes back into the original chunk because they are scattered over temporary chunks */
1351 for (idx = 0, len = ARR_LEN(c->n); idx < len; ++idx) {
1352 const ir_node *n = c->n[idx];
1353 co_mst_irn_t *nn = get_co_mst_irn(env, n);
1357 /* fragment the remaining chunk */
1358 visited = bitset_irg_malloc(env->co->irg);
1359 for (idx = 0, len = ARR_LEN(best_chunk->n); idx < len; ++idx)
1360 bitset_set(visited, get_irn_idx(best_chunk->n[idx]));
1362 for (idx = 0, len = ARR_LEN(c->n); idx < len; ++idx) {
1363 const ir_node *irn = c->n[idx];
1364 if (! bitset_is_set(visited, get_irn_idx(irn))) {
1365 aff_chunk_t *new_chunk = new_aff_chunk(env);
1366 co_mst_irn_t *node = get_co_mst_irn(env, irn);
1368 expand_chunk_from(env, node, visited, new_chunk, c, decider_always_yes, 0);
1369 aff_chunk_assure_weight(env, new_chunk);
1370 pqueue_put(env->chunks, new_chunk, new_chunk->weight);
1374 for (idx = 0, len = ARR_LEN(best_chunk->n); idx < len; ++idx) {
1375 const ir_node *n = best_chunk->n[idx];
1376 co_mst_irn_t *nn = get_co_mst_irn(env, n);
1380 /* clear obsolete chunks and free some memory */
1381 delete_aff_chunk(env, best_chunk);
1382 bitset_free(visited);
1384 del_waitq(best_starts);
1386 stat_ev_ctx_pop("heur4_color_chunk");
1390 * Main driver for mst safe coalescing algorithm.
1392 int co_solve_heuristic_mst(copy_opt_t *co) {
1393 unsigned n_regs = co->cls->n_regs;
1394 bitset_t *ignore_regs = bitset_alloca(n_regs);
1397 co_mst_env_t mst_env;
1404 phase_init(&mst_env.ph, "co_mst", co->irg, PHASE_DEFAULT_GROWTH, co_mst_irn_init, &mst_env);
1406 k = be_put_ignore_regs(co->cenv->birg, co->cls, ignore_regs);
1409 mst_env.n_regs = n_regs;
1411 mst_env.chunks = new_pqueue();
1413 mst_env.ignore_regs = ignore_regs;
1414 mst_env.ifg = co->cenv->ifg;
1415 mst_env.chunkset = pset_new_ptr(512);
1416 mst_env.chunk_visited = 0;
1417 mst_env.single_cols = phase_alloc(&mst_env.ph, sizeof(*mst_env.single_cols) * n_regs);
1419 for (i = 0; i < n_regs; ++i) {
1420 col_cost_t *vec = phase_alloc(&mst_env.ph, sizeof(*vec) * n_regs);
1422 mst_env.single_cols[i] = vec;
1423 for (j = 0; j < n_regs; ++j) {
1425 vec[j].cost = REAL(0.0);
1429 vec[0].cost = REAL(1.0);
1432 DBG((dbg, LEVEL_1, "==== Coloring %+F, class %s ====\n", co->irg, co->cls->name));
1434 /* build affinity chunks */
1436 build_affinity_chunks(&mst_env);
1437 stat_ev_tim_pop("heur4_initial_chunk");
1439 /* color chunks as long as there are some */
1440 while (! pqueue_empty(mst_env.chunks)) {
1441 aff_chunk_t *chunk = pqueue_pop_front(mst_env.chunks);
1443 color_aff_chunk(&mst_env, chunk);
1444 DB((dbg, LEVEL_4, "<<<====== Coloring chunk (%u) done\n", chunk->id));
1445 delete_aff_chunk(&mst_env, chunk);
1448 /* apply coloring */
1449 foreach_phase_irn(&mst_env.ph, irn) {
1451 const arch_register_t *reg;
1453 if (arch_irn_is(irn, ignore))
1456 mirn = get_co_mst_irn(&mst_env, irn);
1457 // assert(mirn->fixed && "Node should have fixed color");
1459 /* skip nodes where color hasn't changed */
1460 if (mirn->init_col == mirn->col)
1463 reg = arch_register_for_index(co->cls, mirn->col);
1464 arch_set_irn_register(irn, reg);
1465 DB((dbg, LEVEL_1, "%+F set color from %d to %d\n", irn, mirn->init_col, mirn->col));
1468 /* free allocated memory */
1469 del_pqueue(mst_env.chunks);
1470 phase_free(&mst_env.ph);
1471 del_pset(mst_env.chunkset);
1473 stat_ev_tim_pop("heur4_total");
1478 static const lc_opt_table_entry_t options[] = {
1479 LC_OPT_ENT_INT ("limit", "limit recoloring", &recolor_limit),
1480 LC_OPT_ENT_DBL ("di", "dislike influence", &dislike_influence),
1485 void be_init_copyheur4(void) {
1486 lc_opt_entry_t *be_grp = lc_opt_get_grp(firm_opt_get_root(), "be");
1487 lc_opt_entry_t *ra_grp = lc_opt_get_grp(be_grp, "ra");
1488 lc_opt_entry_t *chordal_grp = lc_opt_get_grp(ra_grp, "chordal");
1489 lc_opt_entry_t *co_grp = lc_opt_get_grp(chordal_grp, "co");
1490 lc_opt_entry_t *heur4_grp = lc_opt_get_grp(co_grp, "heur4");
1492 lc_opt_add_table(heur4_grp, options);
1493 FIRM_DBG_REGISTER(dbg, "firm.be.co.heur4");
1497 BE_REGISTER_MODULE_CONSTRUCTOR(be_init_copyheur4);