-/**
- * Author: Christian Wuerdig
- * Date: 2005/12/14
- * Copyright: (c) Universitaet Karlsruhe
- * Licence: This file protected by GPL - GNU GENERAL PUBLIC LICENSE.
- * CVS-Id: $Id$
+/*
+ * Copyright (C) 1995-2011 University of Karlsruhe. All right reserved.
+ *
+ * This file is part of libFirm.
+ *
+ * This file may be distributed and/or modified under the terms of the
+ * GNU General Public License version 2 as published by the Free Software
+ * Foundation and appearing in the file LICENSE.GPL included in the
+ * packaging of this file.
*
- * Performs lowering of perm nodes and spill/reload optimization.
+ * Licensees holding valid libFirm Professional Edition licenses may use
+ * this file in accordance with the libFirm Commercial License.
+ * Agreement provided with the Software.
+ *
+ * This file is provided AS IS with NO WARRANTY OF ANY KIND, INCLUDING THE
+ * WARRANTY OF DESIGN, MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+ * PURPOSE.
+ */
+
+/**
+ * @file
+ * @brief Performs lowering of perm nodes. Inserts copies to assure register constraints.
+ * @author Christian Wuerdig
+ * @date 14.12.2005
+ * @version $Id$
*/
-#ifdef HAVE_CONFIG_H
#include "config.h"
-#endif
-#include "bearch.h"
-#include "belower.h"
-#include "benode_t.h"
-#include "bechordal_t.h"
-#include "besched.h"
+#include <stdlib.h>
+#include "ircons.h"
+#include "debug.h"
+#include "xmalloc.h"
+#include "irnodeset.h"
+#include "irnodemap.h"
#include "irgmod.h"
#include "iredges_t.h"
#include "irgwalk.h"
+#include "array_t.h"
-#undef is_Perm
-#define is_Perm(arch_env, irn) (arch_irn_classify(arch_env, irn) == arch_irn_class_perm)
-
-/* lowering walker environment */
-typedef struct _lower_env_t {
- be_chordal_env_t *chord_env;
- int do_copy;
- firm_dbg_module_t *dbg_module;
+#include "bearch.h"
+#include "belower.h"
+#include "benode.h"
+#include "besched.h"
+#include "bestat.h"
+#include "bessaconstr.h"
+#include "beintlive_t.h"
+
+#undef KEEP_ALIVE_COPYKEEP_HACK
+
+DEBUG_ONLY(static firm_dbg_module_t *dbg;)
+DEBUG_ONLY(static firm_dbg_module_t *dbg_constr;)
+DEBUG_ONLY(static firm_dbg_module_t *dbg_permmove;)
+
+/** Associates an ir_node with its copy and CopyKeep. */
+typedef struct {
+ ir_nodeset_t copies; /**< all non-spillable copies of this irn */
+ const arch_register_class_t *cls;
+} op_copy_assoc_t;
+
+/** Environment for constraints. */
+typedef struct {
+ ir_graph *irg;
+ ir_nodemap_t op_set;
+ struct obstack obst;
+} constraint_env_t;
+
+/** Lowering walker environment. */
+typedef struct lower_env_t {
+ ir_graph *irg;
+ unsigned do_copy : 1;
} lower_env_t;
-/* holds a perm register pair */
-typedef struct _reg_pair_t {
+/** Holds a Perm register pair. */
+typedef struct reg_pair_t {
const arch_register_t *in_reg; /**< a perm IN register */
ir_node *in_node; /**< the in node to which the register belongs */
int checked; /**< indicates whether the pair was check for cycle or not */
} reg_pair_t;
-typedef enum _perm_type_t {
+typedef enum perm_type_t {
PERM_CYCLE,
PERM_CHAIN,
PERM_SWAP,
PERM_COPY
} perm_type_t;
-/* structure to represent cycles or chains in a perm */
-typedef struct _perm_cycle_t {
+/** Structure to represent cycles or chains in a Perm. */
+typedef struct perm_cycle_t {
const arch_register_t **elems; /**< the registers in the cycle */
int n_elems; /**< number of elements in the cycle */
perm_type_t type; /**< type (CHAIN or CYCLE) */
} perm_cycle_t;
-/* Compare the in registers of two register pairs */
-static int compare_reg_pair(const void *a, const void *b) {
- const reg_pair_t *pair_a = a;
- const reg_pair_t *pair_b = b;
+/** returns the number register pairs marked as checked. */
+static int get_n_unchecked_pairs(reg_pair_t const *const pairs, int const n)
+{
+ int n_unchecked = 0;
+ int i;
- if (pair_a->in_reg->index > pair_b->in_reg->index)
- return 1;
- else
- return -1;
+ for (i = 0; i < n; i++) {
+ if (!pairs[i].checked)
+ n_unchecked++;
+ }
+
+ return n_unchecked;
}
-/* returns the number register pairs marked as checked */
-static int get_n_checked_pairs(reg_pair_t *pairs, int n) {
- int i, n_checked = 0;
+/**
+ * Gets the node corresponding to an IN register from an array of register pairs.
+ * NOTE: The given registers pairs and the register to look for must belong
+ * to the same register class.
+ *
+ * @param pairs The array of register pairs
+ * @param n The number of pairs
+ * @param reg The register to look for
+ * @return The corresponding node or NULL if not found
+ */
+static ir_node *get_node_for_in_register(reg_pair_t *pairs, int n, const arch_register_t *reg)
+{
+ int i;
for (i = 0; i < n; i++) {
- if (pairs[i].checked)
- n_checked++;
+ /* in register matches */
+ if (pairs[i].in_reg->index == reg->index)
+ return pairs[i].in_node;
}
- return n_checked;
+ return NULL;
}
/**
- * Gets the node corresponding to a register from an array of register pairs.
+ * Gets the node corresponding to an OUT register from an array of register pairs.
* NOTE: The given registers pairs and the register to look for must belong
* to the same register class.
*
* @param pairs The array of register pairs
* @param n The number of pairs
* @param reg The register to look for
- * @param in_out 0 == look for IN register, 1 == look for OUT register
* @return The corresponding node or NULL if not found
*/
-static ir_node *get_node_for_register(reg_pair_t *pairs, int n, const arch_register_t *reg, int in_out) {
+static ir_node *get_node_for_out_register(reg_pair_t *pairs, int n, const arch_register_t *reg)
+{
int i;
- if (in_out) {
- for (i = 0; i < n; i++) {
- /* out register matches */
- if (pairs[i].out_reg->index == reg->index)
- return pairs[i].out_node;
- }
- }
- else {
- for (i = 0; i < n; i++) {
- /* in register matches */
- if (pairs[i].in_reg->index == reg->index)
- return pairs[i].in_node;
- }
+ for (i = 0; i < n; i++) {
+ /* out register matches */
+ if (pairs[i].out_reg->index == reg->index)
+ return pairs[i].out_node;
}
return NULL;
}
/**
- * Gets the index in the register pair array where the in/out register
+ * Gets the index in the register pair array where the in register
* corresponds to reg_idx.
*
* @param pairs The array of register pairs
* @param n The number of pairs
* @param reg The register index to look for
- * @param in_out 0 == look for IN register, 1 == look for OUT register
+ *
* @return The corresponding index in pairs or -1 if not found
*/
-static int get_pairidx_for_regidx(reg_pair_t *pairs, int n, int reg_idx, int in_out) {
+static int get_pairidx_for_in_regidx(reg_pair_t *pairs, int n, unsigned reg_idx)
+{
int i;
- if (in_out) {
- for (i = 0; i < n; i++) {
- /* out register matches */
- if (pairs[i].out_reg->index == reg_idx)
- return i;
- }
- }
- else {
- for (i = 0; i < n; i++) {
- /* in register matches */
- if (pairs[i].in_reg->index == reg_idx)
- return i;
- }
+ for (i = 0; i < n; i++) {
+ /* in register matches */
+ if (pairs[i].in_reg->index == reg_idx)
+ return i;
}
+ return -1;
+}
+/**
+ * Gets the index in the register pair array where the out register
+ * corresponds to reg_idx.
+ *
+ * @param pairs The array of register pairs
+ * @param n The number of pairs
+ * @param reg The register index to look for
+ *
+ * @return The corresponding index in pairs or -1 if not found
+ */
+static int get_pairidx_for_out_regidx(reg_pair_t *pairs, int n, unsigned reg_idx)
+{
+ int i;
+
+ for (i = 0; i < n; i++) {
+ /* out register matches */
+ if (pairs[i].out_reg->index == reg_idx)
+ return i;
+ }
return -1;
}
/**
- * Gets an array of register pairs and tries to identify a cycle or chain starting
- * at position start.
+ * Gets an array of register pairs and tries to identify a cycle or chain
+ * starting at position start.
+ *
+ * @param cycle Variable to hold the cycle
+ * @param pairs Array of register pairs
+ * @param n length of the pairs array
+ * @param start Index to start
*
- * @param cycle Variable to hold the cycle
- * @param pairs Array of register pairs
- * @param start Index to start
* @return The cycle or chain
*/
-static perm_cycle_t *get_perm_cycle(perm_cycle_t *cycle, reg_pair_t *pairs, int n, int start) {
- int head = pairs[start].in_reg->index;
- int cur_idx = pairs[start].out_reg->index;
- int cur_pair_idx = start;
- int n_pairs_done = get_n_checked_pairs(pairs, n) + 1;
- int idx;
+static void get_perm_cycle(perm_cycle_t *const cycle,
+ reg_pair_t *const pairs,
+ int const n,
+ int start)
+{
+ int head = pairs[start].in_reg->index;
+ int cur_idx = pairs[start].out_reg->index;
+ int const n_pairs_todo = get_n_unchecked_pairs(pairs, n);
+ perm_type_t cycle_tp = PERM_CYCLE;
+ int idx;
+
+ /* We could be right in the middle of a chain, so we need to find the start */
+ while (head != cur_idx) {
+ /* goto previous register in cycle or chain */
+ int const cur_pair_idx = get_pairidx_for_out_regidx(pairs, n, head);
+
+ if (cur_pair_idx < 0) {
+ cycle_tp = PERM_CHAIN;
+ break;
+ } else {
+ head = pairs[cur_pair_idx].in_reg->index;
+ start = cur_pair_idx;
+ }
+ }
/* assume worst case: all remaining pairs build a cycle or chain */
- cycle->elems = calloc(n - n_pairs_done, sizeof(cycle->elems[0]));
+ cycle->elems = XMALLOCNZ(const arch_register_t*, n_pairs_todo * 2);
cycle->n_elems = 2; /* initial number of elements is 2 */
cycle->elems[0] = pairs[start].in_reg;
cycle->elems[1] = pairs[start].out_reg;
- cycle->type = PERM_CHAIN; /* default is CHAIN, only changed when we found a cycle */
-
- /* mark the first pair as checked */
- pairs[start].checked = 1;
+ cycle->type = cycle_tp;
+ cur_idx = pairs[start].out_reg->index;
idx = 2;
/* check for cycle or end of a chain */
- while (cur_idx != head && n_pairs_done < n) {
+ while (cur_idx != head) {
/* goto next register in cycle or chain */
- cur_pair_idx = get_pairidx_for_regidx(pairs, n, cur_idx, 0);
+ int const cur_pair_idx = get_pairidx_for_in_regidx(pairs, n, cur_idx);
if (cur_pair_idx < 0)
break;
if (cur_idx != head) {
cycle->elems[idx++] = pairs[cur_pair_idx].out_reg;
cycle->n_elems++;
-
- n_pairs_done++;
- }
- else {
+ } else {
/* we are there where we started -> CYCLE */
cycle->type = PERM_CYCLE;
}
-
- /* mark the pair as checked */
- pairs[cur_pair_idx].checked = 1;
}
- return cycle;
+ /* mark all pairs having one in/out register with cycle in common as checked */
+ for (idx = 0; idx < cycle->n_elems; idx++) {
+ int cur_pair_idx;
+
+ cur_pair_idx = get_pairidx_for_in_regidx(pairs, n, cycle->elems[idx]->index);
+ if (cur_pair_idx >= 0)
+ pairs[cur_pair_idx].checked = 1;
+
+ cur_pair_idx = get_pairidx_for_out_regidx(pairs, n, cycle->elems[idx]->index);
+ if (cur_pair_idx >= 0)
+ pairs[cur_pair_idx].checked = 1;
+ }
}
/**
* Lowers a perm node. Resolves cycles and creates a bunch of
* copy and swap operations to permute registers.
+ * Note: The caller of this function has to make sure, that irn
+ * is a Perm node.
*
* @param irn The perm node
* @param block The block the perm node belongs to
- * @param walk_env The environment
+ * @param env The lowerer environment
*/
-static void lower_perm_node(ir_node *irn, void *walk_env) {
- const be_node_factory_t *fact;
- const arch_register_class_t *reg_class;
- const arch_env_t *arch_env;
- firm_dbg_module_t *mod;
- lower_env_t *env = walk_env;
- reg_pair_t *pairs;
- const ir_edge_t *edge;
- perm_cycle_t *cycle;
- int n, i, pn, do_copy;
- ir_node *sched_point, *block, *in[2];
- ir_node *arg1, *arg2, *res1, *res2;
- ir_node *cpyxchg = NULL;
-
- if (is_Block(irn))
- return;
+static void lower_perm_node(ir_node *irn, lower_env_t *env)
+{
+ const arch_register_class_t *const reg_class = arch_get_irn_register(get_irn_n(irn, 0))->reg_class;
+ ir_node *const block = get_nodes_block(irn);
+ int const arity = get_irn_arity(irn);
+ reg_pair_t *const pairs = ALLOCAN(reg_pair_t, arity);
+ int keep_perm = 0;
+ int do_copy = env->do_copy;
+ /* Get the schedule predecessor node to the perm.
+ * NOTE: This works with auto-magic. If we insert the new copy/exchange
+ * nodes after this node, everything should be ok. */
+ ir_node * sched_point = sched_prev(irn);
+ const ir_edge_t * edge;
+ const ir_edge_t * next;
+ int n;
+ int i;
+
+ DBG((dbg, LEVEL_1, "perm: %+F, sched point is %+F\n", irn, sched_point));
+ assert(sched_point && "Perm is not scheduled or has no predecessor");
- fact = env->chord_env->main_env->node_factory;
- arch_env = env->chord_env->main_env->arch_env;
- do_copy = env->do_copy;
- mod = env->dbg_module;
- block = get_nodes_block(irn);
+ assert(arity == get_irn_n_edges(irn) && "perm's in and out numbers different");
- /* check if perm
- Note: A Proj on a Perm will cause is_Perm()
- also to return true */
- if (is_Proj(irn) || ! is_Perm(arch_env, irn))
- return;
+ /* build the list of register pairs (in, out) */
+ n = 0;
+ foreach_out_edge_safe(irn, edge, next) {
+ ir_node *const out = get_edge_src_irn(edge);
+ long const pn = get_Proj_proj(out);
+ ir_node *const in = get_irn_n(irn, pn);
+ arch_register_t const *const in_reg = arch_get_irn_register(in);
+ arch_register_t const *const out_reg = arch_get_irn_register(out);
+ reg_pair_t * pair;
+
+ if (in_reg == out_reg) {
+ DBG((dbg, LEVEL_1, "%+F removing equal perm register pair (%+F, %+F, %s)\n",
+ irn, in, out, out_reg->name));
+ exchange(out, in);
+ continue;
+ }
- /*
- Get the schedule predecessor node to the perm
- NOTE: This works with auto-magic. If we insert the
- new copy/exchange nodes after this node, everything
- should be ok.
- */
- sched_point = sched_prev(irn);
- DBG((mod, LEVEL_1, "sched point is %+F\n", sched_point));
- assert(sched_point && "Perm is not scheduled or has no predecessor");
+ pair = &pairs[n++];
+ pair->in_node = in;
+ pair->in_reg = in_reg;
+ pair->out_node = out;
+ pair->out_reg = out_reg;
+ pair->checked = 0;
+ }
- n = get_irn_arity(irn);
- assert(n == get_irn_n_edges(irn) && "perm's in and out numbers different");
+ DBG((dbg, LEVEL_1, "%+F has %d unresolved constraints\n", irn, n));
- reg_class = arch_get_irn_register(arch_env, get_irn_n(irn, 0))->reg_class;
- pairs = calloc(n, sizeof(pairs[0]));
+ /* Set do_copy to 0 if it's on but we have no free register */
+ /* TODO check for free register */
+ if (do_copy) {
+ do_copy = 0;
+ }
- /* build the list of register pairs (in, out) */
- i = 0;
- foreach_out_edge(irn, edge) {
- pairs[i].out_node = get_edge_src_irn(edge);
- pn = get_Proj_proj(pairs[i].out_node);
- pairs[i].in_node = get_irn_n(irn, pn);
+ /* check for cycles and chains */
+ while (get_n_unchecked_pairs(pairs, n) > 0) {
+ perm_cycle_t cycle;
+ int j;
- pairs[i].in_reg = arch_get_irn_register(arch_env, pairs[i].in_node);
- pairs[i].out_reg = arch_get_irn_register(arch_env, pairs[i].out_node);
+ /* go to the first not-checked pair */
+ for (i = 0; pairs[i].checked; ++i) {}
+ get_perm_cycle(&cycle, pairs, n, i);
- pairs[i].checked = 0;
- i++;
+ DB((dbg, LEVEL_1, "%+F: following %s created:\n ", irn, cycle.type == PERM_CHAIN ? "chain" : "cycle"));
+ for (j = 0; j < cycle.n_elems; j++) {
+ DB((dbg, LEVEL_1, " %s", cycle.elems[j]->name));
+ }
+ DB((dbg, LEVEL_1, "\n"));
+
+ if (cycle.type == PERM_CYCLE && arity == 2) {
+ /* We don't need to do anything if we have a Perm with two elements
+ * which represents a cycle, because those nodes already represent
+ * exchange nodes */
+ keep_perm = 1;
+ } else {
+ /* TODO: - iff PERM_CYCLE && do_copy -> determine free temp reg and
+ * insert copy to/from it before/after the copy cascade (this
+ * reduces the cycle into a chain) */
+
+ /* build copy/swap nodes from back to front */
+ for (i = cycle.n_elems - 2; i >= 0; i--) {
+ ir_node *arg1 = get_node_for_in_register(pairs, n, cycle.elems[i]);
+ ir_node *arg2 = get_node_for_in_register(pairs, n, cycle.elems[i + 1]);
+
+ ir_node *res1 = get_node_for_out_register(pairs, n, cycle.elems[i]);
+ ir_node *res2 = get_node_for_out_register(pairs, n, cycle.elems[i + 1]);
+ /* If we have a cycle and don't copy: we need to create exchange
+ * nodes
+ * NOTE: An exchange node is a perm node with 2 INs and 2 OUTs
+ * IN_1 = in node with register i
+ * IN_2 = in node with register i + 1
+ * OUT_1 = out node with register i + 1
+ * OUT_2 = out node with register i */
+ if (cycle.type == PERM_CYCLE && !do_copy) {
+ ir_node *in[2];
+ ir_node *cpyxchg;
+
+ in[0] = arg1;
+ in[1] = arg2;
+
+ /* At this point we have to handle the following problem:
+ *
+ * If we have a cycle with more than two elements, then this
+ * could correspond to the following Perm node:
+ *
+ * +----+ +----+ +----+
+ * | r1 | | r2 | | r3 |
+ * +-+--+ +-+--+ +--+-+
+ * | | |
+ * | | |
+ * +-+--------+---------+-+
+ * | Perm |
+ * +-+--------+---------+-+
+ * | | |
+ * | | |
+ * +-+--+ +-+--+ +--+-+
+ * |Proj| |Proj| |Proj|
+ * | r2 | | r3 | | r1 |
+ * +----+ +----+ +----+
+ *
+ * This node is about to be split up into two 2x Perm's for
+ * which we need 4 Proj's and the one additional Proj of the
+ * first Perm has to be one IN of the second. So in general
+ * we need to create one additional Proj for each "middle"
+ * Perm and set this to one in node of the successor Perm. */
+
+ DBG((dbg, LEVEL_1, "%+F creating exchange node (%+F, %s) and (%+F, %s) with\n",
+ irn, arg1, cycle.elems[i]->name, arg2, cycle.elems[i + 1]->name));
+ DBG((dbg, LEVEL_1, "%+F (%+F, %s) and (%+F, %s)\n",
+ irn, res1, cycle.elems[i]->name, res2, cycle.elems[i + 1]->name));
+
+ cpyxchg = be_new_Perm(reg_class, block, 2, in);
+
+ if (i > 0) {
+ /* cycle is not done yet */
+ int pidx = get_pairidx_for_in_regidx(pairs, n, cycle.elems[i]->index);
+
+ /* create intermediate proj */
+ res1 = new_r_Proj(cpyxchg, get_irn_mode(res1), 0);
+
+ /* set as in for next Perm */
+ pairs[pidx].in_node = res1;
+ }
+
+ set_Proj_pred(res2, cpyxchg);
+ set_Proj_proj(res2, 0);
+ set_Proj_pred(res1, cpyxchg);
+ set_Proj_proj(res1, 1);
+
+ arch_set_irn_register(res2, cycle.elems[i + 1]);
+ arch_set_irn_register(res1, cycle.elems[i]);
+
+ /* insert the copy/exchange node in schedule after the magic schedule node (see above) */
+ sched_add_after(skip_Proj(sched_point), cpyxchg);
+
+ DB((dbg, LEVEL_1, "replacing %+F with %+F, placed new node after %+F\n", irn, cpyxchg, sched_point));
+
+ /* set the new scheduling point */
+ sched_point = res1;
+ } else {
+ ir_node *cpyxchg;
+
+ DB((dbg, LEVEL_1, "%+F creating copy node (%+F, %s) -> (%+F, %s)\n",
+ irn, arg1, cycle.elems[i]->name, res2, cycle.elems[i + 1]->name));
+
+ cpyxchg = be_new_Copy(block, arg1);
+ arch_set_irn_register(cpyxchg, cycle.elems[i + 1]);
+
+ /* exchange copy node and proj */
+ exchange(res2, cpyxchg);
+
+ /* insert the copy/exchange node in schedule after the magic schedule node (see above) */
+ sched_add_after(skip_Proj(sched_point), cpyxchg);
+
+ /* set the new scheduling point */
+ sched_point = cpyxchg;
+ }
+ }
+ }
+
+ free((void*)cycle.elems);
}
- /* sort the register pairs by the indices of the in registers */
- qsort(pairs, n, sizeof(pairs[0]), compare_reg_pair);
+ /* remove the perm from schedule */
+ if (!keep_perm) {
+ sched_remove(irn);
+ kill_node(irn);
+ }
+}
- /* Mark all equal pairs as checked, and exchange the OUT proj with
- the IN node. */
- for (i = 0; i < n; i++) {
- if (pairs[i].in_reg->index == pairs[i].out_reg->index) {
- DBG((mod, LEVEL_1, "%+F removing equal perm register pair (%+F, %+F, %s)\n",
- irn, pairs[i].in_node, pairs[i].out_node, pairs[i].out_reg->name));
-
- /* We have to check for a special case:
- The in-node could be a Proj from a Perm. In this case,
- we need to correct the projnum */
- if (is_Perm(arch_env, pairs[i].in_node) && is_Proj(pairs[i].in_node)) {
- set_Proj_proj(pairs[i].out_node, get_Proj_proj(pairs[i].in_node));
- }
- /* remove the proj from the schedule */
- sched_remove(pairs[i].out_node);
- /* reroute the edges from the proj to the argument */
- edges_reroute(pairs[i].out_node, pairs[i].in_node, env->chord_env->irg);
+static int has_irn_users(const ir_node *irn)
+{
+ return get_irn_out_edge_first_kind(irn, EDGE_KIND_NORMAL) != 0;
+}
- pairs[i].checked = 1;
- }
+static ir_node *find_copy(ir_node *irn, ir_node *op)
+{
+ ir_node *cur_node;
+
+ for (cur_node = irn;;) {
+ cur_node = sched_prev(cur_node);
+ if (! be_is_Copy(cur_node))
+ return NULL;
+ if (be_get_Copy_op(cur_node) == op && arch_irn_is(cur_node, dont_spill))
+ return cur_node;
+ }
+}
+
+static void gen_assure_different_pattern(ir_node *irn, ir_node *other_different, constraint_env_t *env)
+{
+ ir_nodemap_t *op_set;
+ ir_node *block;
+ const arch_register_class_t *cls;
+ ir_node *keep, *cpy;
+ op_copy_assoc_t *entry;
+
+ if (arch_irn_is_ignore(other_different) ||
+ !mode_is_datab(get_irn_mode(other_different))) {
+ DB((dbg_constr, LEVEL_1, "ignore constraint for %+F because other_irn is ignore or not a datab node\n", irn));
+ return;
}
- /* Set do_copy to 0 if it's on but we have no free register */
- if (do_copy) {
- do_copy = 0;
+ op_set = &env->op_set;
+ block = get_nodes_block(irn);
+ cls = arch_get_irn_reg_class(other_different);
+
+ /* Make a not spillable copy of the different node */
+ /* this is needed because the different irn could be */
+ /* in block far far away */
+ /* The copy is optimized later if not needed */
+
+ /* check if already exists such a copy in the schedule immediately before */
+ cpy = find_copy(skip_Proj(irn), other_different);
+ if (! cpy) {
+ cpy = be_new_Copy(block, other_different);
+ arch_set_irn_flags(cpy, arch_irn_flags_dont_spill);
+ DB((dbg_constr, LEVEL_1, "created non-spillable %+F for value %+F\n", cpy, other_different));
+ } else {
+ DB((dbg_constr, LEVEL_1, "using already existing %+F for value %+F\n", cpy, other_different));
}
- /* check for cycles and chains */
- while (get_n_checked_pairs(pairs, n) < n) {
- i = 0;
+ /* Add the Keep resp. CopyKeep and reroute the users */
+ /* of the other_different irn in case of CopyKeep. */
+ if (has_irn_users(other_different)) {
+ keep = be_new_CopyKeep_single(block, cpy, irn);
+ be_node_set_reg_class_in(keep, 1, cls);
+ } else {
+ ir_node *in[2];
+
+ in[0] = irn;
+ in[1] = cpy;
+ keep = be_new_Keep(block, 2, in);
+ }
- /* go to the first not-checked pair */
- while (pairs[i].checked) i++;
- cycle = calloc(1, sizeof(*cycle));
- cycle = get_perm_cycle(cycle, pairs, n, i);
-
- /* We don't need to do anything if we have a Perm with two
- elements which represents a cycle, because those nodes
- already represent exchange nodes */
- if (n == 2 && cycle->type == PERM_CYCLE) {
- free(cycle);
- continue;
+ DB((dbg_constr, LEVEL_1, "created %+F(%+F, %+F)\n\n", keep, irn, cpy));
+
+ /* insert copy and keep into schedule */
+ assert(sched_is_scheduled(irn) && "need schedule to assure constraints");
+ if (! sched_is_scheduled(cpy))
+ sched_add_before(skip_Proj(irn), cpy);
+ sched_add_after(skip_Proj(irn), keep);
+
+ /* insert the other different and its copies into the map */
+ entry = (op_copy_assoc_t*)ir_nodemap_get(op_set, other_different);
+ if (! entry) {
+ entry = OALLOC(&env->obst, op_copy_assoc_t);
+ entry->cls = cls;
+ ir_nodeset_init(&entry->copies);
+
+ ir_nodemap_insert(op_set, other_different, entry);
+ }
+
+ /* insert copy */
+ ir_nodeset_insert(&entry->copies, cpy);
+
+ /* insert keep in case of CopyKeep */
+ if (be_is_CopyKeep(keep))
+ ir_nodeset_insert(&entry->copies, keep);
+}
+
+/**
+ * Checks if node has a must_be_different constraint in output and adds a Keep
+ * then to assure the constraint.
+ *
+ * @param irn the node to check
+ * @param skipped_irn if irn is a Proj node, its predecessor, else irn
+ * @param env the constraint environment
+ */
+static void assure_different_constraints(ir_node *irn, ir_node *skipped_irn, constraint_env_t *env)
+{
+ const arch_register_req_t *req = arch_get_irn_register_req(irn);
+
+ if (arch_register_req_is(req, must_be_different)) {
+ const unsigned other = req->other_different;
+ int i;
+
+ if (arch_register_req_is(req, should_be_same)) {
+ const unsigned same = req->other_same;
+
+ if (is_po2(other) && is_po2(same)) {
+ int idx_other = ntz(other);
+ int idx_same = ntz(same);
+
+ /*
+ * We can safely ignore a should_be_same x must_be_different y
+ * IFF both inputs are equal!
+ */
+ if (get_irn_n(skipped_irn, idx_other) == get_irn_n(skipped_irn, idx_same)) {
+ return;
+ }
+ }
+ }
+ for (i = 0; 1U << i <= other; ++i) {
+ if (other & (1U << i)) {
+ ir_node *different_from = get_irn_n(skipped_irn, i);
+ gen_assure_different_pattern(irn, different_from, env);
+ }
}
+ }
+}
+
+/**
+ * Calls the functions to assure register constraints.
+ *
+ * @param block The block to be checked
+ * @param walk_env The walker environment
+ */
+static void assure_constraints_walker(ir_node *block, void *walk_env)
+{
+ ir_node *irn;
+ constraint_env_t *env = (constraint_env_t*)walk_env;
-//TODO: - iff PERM_CYCLE && do_copy -> determine free temp reg and insert copy to/from it before/after
-// the copy cascade (this reduces the cycle into a chain)
-
- /* build copy/swap nodes from back to front */
- for (i = cycle->n_elems - 2; i >= 0; i--) {
- arg1 = get_node_for_register(pairs, n, cycle->elems[i], 0);
- arg2 = get_node_for_register(pairs, n, cycle->elems[i + 1], 0);
-
- res1 = get_node_for_register(pairs, n, cycle->elems[i], 1);
- res2 = get_node_for_register(pairs, n, cycle->elems[i + 1], 1);
-
- /*
- If we have a cycle and don't copy: we need to create exchange nodes
- NOTE: An exchange node is a perm node with 2 INs and 2 OUTs
- IN_1 = in node with register i
- IN_2 = in node with register i + 1
- OUT_1 = out node with register i + 1
- OUT_2 = out node with register i
- */
- if (cycle->type == PERM_CYCLE && !do_copy) {
- in[0] = arg1;
- in[1] = arg2;
-
- DBG((mod, LEVEL_1, "%+F creating exchange node (%+F, %s) and (%+F, %s) with\n",
- irn, arg1, cycle->elems[i]->name, arg2, cycle->elems[i + 1]->name));
- DBG((mod, LEVEL_1, "%+F (%+F, %s) and (%+F, %s)\n",
- irn, res1, cycle->elems[i]->name, res2, cycle->elems[i + 1]->name));
-
- cpyxchg = new_Perm(fact, reg_class, env->chord_env->irg, block, 2, in);
-
- sched_remove(res1);
- sched_remove(res2);
-
- set_Proj_pred(res2, cpyxchg);
- set_Proj_proj(res2, 0);
- set_Proj_pred(res1, cpyxchg);
- set_Proj_proj(res1, 1);
-
- sched_add_after(sched_point, res1);
- sched_add_after(sched_point, res2);
-
- arch_set_irn_register(arch_env, res2, cycle->elems[i + 1]);
- arch_set_irn_register(arch_env, res1, cycle->elems[i]);
+ sched_foreach_reverse(block, irn) {
+ ir_mode *mode = get_irn_mode(irn);
+
+ if (mode == mode_T) {
+ const ir_edge_t *edge;
+
+ foreach_out_edge(irn, edge) {
+ ir_node *proj = get_edge_src_irn(edge);
+
+ mode = get_irn_mode(proj);
+ if (mode_is_datab(mode))
+ assure_different_constraints(proj, irn, env);
}
+ } else if (mode_is_datab(mode)) {
+ assure_different_constraints(irn, irn, env);
+ }
+ }
+}
+
+/**
+ * Melt all copykeeps pointing to the same node
+ * (or Projs of the same node), copying the same operand.
+ */
+static void melt_copykeeps(constraint_env_t *cenv)
+{
+ ir_nodemap_iterator_t map_iter;
+ ir_nodemap_entry_t map_entry;
+
+ /* for all */
+ foreach_ir_nodemap(&cenv->op_set, map_entry, map_iter) {
+ op_copy_assoc_t *entry = (op_copy_assoc_t*)map_entry.data;
+ int idx, num_ck;
+ ir_node *cp;
+ struct obstack obst;
+ ir_nodeset_iterator_t iter;
+ ir_node **ck_arr, **melt_arr;
+
+ obstack_init(&obst);
+
+ /* collect all copykeeps */
+ num_ck = idx = 0;
+ foreach_ir_nodeset(&entry->copies, cp, iter) {
+ if (be_is_CopyKeep(cp)) {
+ obstack_grow(&obst, &cp, sizeof(cp));
+ ++num_ck;
+ }
+#ifdef KEEP_ALIVE_COPYKEEP_HACK
else {
- DBG((mod, LEVEL_1, "%+F creating copy node (%+F, %s) -> (%+F, %s)\n",
- irn, arg1, cycle->elems[i]->name, res2, cycle->elems[i + 1]->name));
+ set_irn_mode(cp, mode_ANY);
+ keep_alive(cp);
+ }
+#endif /* KEEP_ALIVE_COPYKEEP_HACK */
+ }
- cpyxchg = new_Copy(fact, reg_class, env->chord_env->irg, block, arg1);
- arch_set_irn_register(arch_env, cpyxchg, cycle->elems[i + 1]);
+ /* compare each copykeep with all other copykeeps */
+ ck_arr = (ir_node **)obstack_finish(&obst);
+ for (idx = 0; idx < num_ck; ++idx) {
+ ir_node *ref, *ref_mode_T;
+
+ if (ck_arr[idx]) {
+ int j, n_melt;
+ ir_node **new_ck_in;
+ ir_node *new_ck;
+ ir_node *sched_pt = NULL;
+
+ n_melt = 1;
+ ref = ck_arr[idx];
+ ref_mode_T = skip_Proj(get_irn_n(ref, 1));
+ obstack_grow(&obst, &ref, sizeof(ref));
+
+ DB((dbg_constr, LEVEL_1, "Trying to melt %+F:\n", ref));
+
+ /* check for copykeeps pointing to the same mode_T node as the reference copykeep */
+ for (j = 0; j < num_ck; ++j) {
+ ir_node *cur_ck = ck_arr[j];
+
+ if (j != idx && cur_ck && skip_Proj(get_irn_n(cur_ck, 1)) == ref_mode_T) {
+ obstack_grow(&obst, &cur_ck, sizeof(cur_ck));
+ ir_nodeset_remove(&entry->copies, cur_ck);
+ DB((dbg_constr, LEVEL_1, "\t%+F\n", cur_ck));
+ ck_arr[j] = NULL;
+ ++n_melt;
+ sched_remove(cur_ck);
+ }
+ }
+ ck_arr[idx] = NULL;
+
+ /* check, if we found some candidates for melting */
+ if (n_melt == 1) {
+ DB((dbg_constr, LEVEL_1, "\tno candidate found\n"));
+ continue;
+ }
+
+ ir_nodeset_remove(&entry->copies, ref);
+ sched_remove(ref);
+
+ melt_arr = (ir_node **)obstack_finish(&obst);
+ /* melt all found copykeeps */
+ NEW_ARR_A(ir_node *, new_ck_in, n_melt);
+ for (j = 0; j < n_melt; ++j) {
+ new_ck_in[j] = get_irn_n(melt_arr[j], 1);
+
+ /* now, we can kill the melted keep, except the */
+ /* ref one, we still need some information */
+ if (melt_arr[j] != ref)
+ kill_node(melt_arr[j]);
+ }
+
+#ifdef KEEP_ALIVE_COPYKEEP_HACK
+ new_ck = be_new_CopyKeep(get_nodes_block(ref), be_get_CopyKeep_op(ref), n_melt, new_ck_in);
+ keep_alive(new_ck);
+#else
+ new_ck = be_new_CopyKeep(get_nodes_block(ref), be_get_CopyKeep_op(ref), n_melt, new_ck_in);
+#endif /* KEEP_ALIVE_COPYKEEP_HACK */
+
+ /* set register class for all kept inputs */
+ for (j = 1; j <= n_melt; ++j)
+ be_node_set_reg_class_in(new_ck, j, entry->cls);
+
+ ir_nodeset_insert(&entry->copies, new_ck);
+
+ /* find scheduling point */
+ sched_pt = ref_mode_T;
+ do {
+ /* just walk along the schedule until a non-Keep/CopyKeep node is found */
+ sched_pt = sched_next(sched_pt);
+ } while (be_is_Keep(sched_pt) || be_is_CopyKeep(sched_pt));
+
+ sched_add_before(sched_pt, new_ck);
+ DB((dbg_constr, LEVEL_1, "created %+F, scheduled before %+F\n", new_ck, sched_pt));
+
+ /* finally: kill the reference copykeep */
+ kill_node(ref);
+ }
+ }
- /* remove the proj from the schedule */
- sched_remove(res2);
+ obstack_free(&obst, NULL);
+ }
+}
- /* exchange copy node and proj */
- exchange(res2, cpyxchg);
+void assure_constraints(ir_graph *irg)
+{
+ constraint_env_t cenv;
+ ir_nodemap_iterator_t map_iter;
+ ir_nodemap_entry_t map_entry;
+
+ FIRM_DBG_REGISTER(dbg_constr, "firm.be.lower.constr");
+
+ cenv.irg = irg;
+ ir_nodemap_init(&cenv.op_set);
+ obstack_init(&cenv.obst);
+
+ irg_block_walk_graph(irg, NULL, assure_constraints_walker, &cenv);
+
+ /* melt copykeeps, pointing to projs of */
+ /* the same mode_T node and keeping the */
+ /* same operand */
+ melt_copykeeps(&cenv);
+
+ /* for all */
+ foreach_ir_nodemap(&cenv.op_set, map_entry, map_iter) {
+ op_copy_assoc_t *entry = (op_copy_assoc_t*)map_entry.data;
+ size_t n = ir_nodeset_size(&entry->copies);
+ ir_node **nodes = ALLOCAN(ir_node*, n);
+ ir_node *cp;
+ ir_nodeset_iterator_t iter;
+ be_ssa_construction_env_t senv;
+
+ /* put the node in an array */
+ DBG((dbg_constr, LEVEL_1, "introduce copies for %+F ", map_entry.node));
+
+ /* collect all copies */
+ n = 0;
+ foreach_ir_nodeset(&entry->copies, cp, iter) {
+ nodes[n++] = cp;
+ DB((dbg_constr, LEVEL_1, ", %+F ", cp));
+ }
+
+ DB((dbg_constr, LEVEL_1, "\n"));
+
+ /* introduce the copies for the operand and its copies */
+ be_ssa_construction_init(&senv, irg);
+ be_ssa_construction_add_copy(&senv, map_entry.node);
+ be_ssa_construction_add_copies(&senv, nodes, n);
+ be_ssa_construction_fix_users(&senv, map_entry.node);
+ be_ssa_construction_destroy(&senv);
+
+ /* Could be that not all CopyKeeps are really needed, */
+ /* so we transform unnecessary ones into Keeps. */
+ foreach_ir_nodeset(&entry->copies, cp, iter) {
+ if (be_is_CopyKeep(cp) && get_irn_n_edges(cp) < 1) {
+ int n = get_irn_arity(cp);
+ ir_node *keep;
+
+ keep = be_new_Keep(get_nodes_block(cp), n, get_irn_in(cp) + 1);
+ sched_add_before(cp, keep);
+
+ /* Set all ins (including the block) of the CopyKeep BAD to keep the verifier happy. */
+ sched_remove(cp);
+ kill_node(cp);
}
+ }
- /* insert the copy/exchange node in schedule after the magic schedule node (see above) */
- sched_add_after(sched_point, cpyxchg);
+ ir_nodeset_destroy(&entry->copies);
+ }
+
+ ir_nodemap_destroy(&cenv.op_set);
+ obstack_free(&cenv.obst, NULL);
+ be_liveness_invalidate(be_get_irg_liveness(irg));
+}
- DBG((mod, LEVEL_1, "replacing %+F with %+F, placed new node after %+F\n", irn, cpyxchg, sched_point));
+
+/**
+ * Push nodes that do not need to be permed through the Perm.
+ * This is commonly a reload cascade at block ends.
+ * @note This routine needs interference.
+ * @note Probably, we can implement it a little more efficient.
+ * Especially searching the frontier lazily might be better.
+ *
+ * @param perm The perm
+ * @param env The lowerer environment
+ *
+ * @return 1, if there is something left to perm over.
+ * 0, if removed the complete perm.
+ */
+static int push_through_perm(ir_node *perm)
+{
+ ir_graph *irg = get_irn_irg(perm);
+ ir_node *bl = get_nodes_block(perm);
+ ir_node *node;
+ int arity = get_irn_arity(perm);
+ int *map;
+ int *proj_map;
+ bitset_t *moved = bitset_alloca(arity);
+ int n_moved;
+ int new_size;
+ ir_node *frontier = bl;
+ ir_node *irn;
+ int i, n;
+
+ /* get some Proj and find out the register class of that Proj. */
+ const ir_edge_t *edge = get_irn_out_edge_first_kind(perm, EDGE_KIND_NORMAL);
+ ir_node *one_proj = get_edge_src_irn(edge);
+ const arch_register_class_t *cls = arch_get_irn_reg_class(one_proj);
+ assert(is_Proj(one_proj));
+
+ DB((dbg_permmove, LEVEL_1, "perm move %+F irg %+F\n", perm, irg));
+
+ /* Find the point in the schedule after which the
+ * potentially movable nodes must be defined.
+ * A Perm will only be pushed up to first instruction
+ * which lets an operand of itself die.
+ * If we would allow to move the Perm above this instruction,
+ * the former dead operand would be live now at the point of
+ * the Perm, increasing the register pressure by one.
+ */
+ sched_foreach_reverse_from(sched_prev(perm), irn) {
+ for (i = get_irn_arity(irn) - 1; i >= 0; --i) {
+ ir_node *op = get_irn_n(irn, i);
+ be_lv_t *lv = be_get_irg_liveness(irg);
+ if (arch_irn_consider_in_reg_alloc(cls, op) &&
+ !be_values_interfere(lv, op, one_proj)) {
+ frontier = irn;
+ goto found_front;
+ }
+ }
+ }
+found_front:
+
+ DB((dbg_permmove, LEVEL_2, "\tfrontier: %+F\n", frontier));
+
+ node = sched_prev(perm);
+ n_moved = 0;
+ while (!sched_is_begin(node)) {
+ const arch_register_req_t *req;
+ int input = -1;
+ ir_node *proj = NULL;
+
+ /* search if node is a INPUT of Perm */
+ foreach_out_edge(perm, edge) {
+ ir_node *out = get_edge_src_irn(edge);
+ int pn = get_Proj_proj(out);
+ ir_node *in = get_irn_n(perm, pn);
+ if (node == in) {
+ proj = out;
+ input = pn;
+ break;
+ }
+ }
+ /* it wasn't an input to the perm, we can't do anything more */
+ if (input < 0)
+ break;
+ if (!sched_comes_after(frontier, node))
+ break;
+ if (arch_irn_is(node, modify_flags))
+ break;
+ req = arch_get_irn_register_req(node);
+ if (req->type != arch_register_req_type_normal)
+ break;
+ for (i = get_irn_arity(node) - 1; i >= 0; --i) {
+ ir_node *opop = get_irn_n(node, i);
+ if (arch_irn_consider_in_reg_alloc(cls, opop)) {
+ break;
+ }
}
+ if (i >= 0)
+ break;
+
+ DBG((dbg_permmove, LEVEL_2, "\tmoving %+F after %+F, killing %+F\n", node, perm, proj));
+
+ /* move the movable node in front of the Perm */
+ sched_remove(node);
+ sched_add_after(perm, node);
+
+ /* give it the proj's register */
+ arch_set_irn_register(node, arch_get_irn_register(proj));
+
+ /* reroute all users of the proj to the moved node. */
+ exchange(proj, node);
-// free(cycle->elems);
- free(cycle);
+ bitset_set(moved, input);
+ n_moved++;
+
+ node = sched_prev(node);
}
- /* remove the perm from schedule */
- sched_remove(irn);
+ /* well, we could not push anything through the perm */
+ if (n_moved == 0)
+ return 1;
+
+ new_size = arity - n_moved;
+ if (new_size == 0) {
+ sched_remove(perm);
+ kill_node(perm);
+ return 0;
+ }
+
+ map = ALLOCAN(int, new_size);
+ proj_map = ALLOCAN(int, arity);
+ memset(proj_map, -1, sizeof(proj_map[0]));
+ n = 0;
+ for (i = 0; i < arity; ++i) {
+ if (bitset_is_set(moved, i))
+ continue;
+ map[n] = i;
+ proj_map[i] = n;
+ n++;
+ }
+ assert(n == new_size);
+ foreach_out_edge(perm, edge) {
+ ir_node *proj = get_edge_src_irn(edge);
+ int pn = get_Proj_proj(proj);
+ pn = proj_map[pn];
+ assert(pn >= 0);
+ set_Proj_proj(proj, pn);
+ }
+
+ be_Perm_reduce(perm, new_size, map);
+ return 1;
}
/**
- * Walks over all blocks in an irg and performs perm lowering.
+ * Calls the corresponding lowering function for the node.
*
- * @param chord_env The chordal environment containing the irg
- * @param do_copy 1 == resolve cycles with a free reg if available
+ * @param irn The node to be checked for lowering
+ * @param walk_env The walker environment
*/
-void lower_perms(be_chordal_env_t *chord_env, int do_copy) {
+static void lower_nodes_after_ra_walker(ir_node *irn, void *walk_env)
+{
+ int perm_stayed;
+
+ if (!be_is_Perm(irn))
+ return;
+
+ perm_stayed = push_through_perm(irn);
+ if (perm_stayed)
+ lower_perm_node(irn, (lower_env_t*)walk_env);
+}
+
+void lower_nodes_after_ra(ir_graph *irg, int do_copy)
+{
lower_env_t env;
- env.chord_env = chord_env;
- env.do_copy = do_copy;
- env.dbg_module = firm_dbg_register("ir.be.lower");
+ FIRM_DBG_REGISTER(dbg, "firm.be.lower");
+ FIRM_DBG_REGISTER(dbg_permmove, "firm.be.lower.permmove");
+
+ env.irg = irg;
+ env.do_copy = do_copy;
+
+ /* we will need interference */
+ be_liveness_assure_chk(be_get_irg_liveness(irg));
- irg_walk_blkwise_graph(chord_env->irg, NULL, lower_perm_node, &env);
+ irg_walk_graph(irg, NULL, lower_nodes_after_ra_walker, &env);
}