1 /*
2 * Copyright (c) 1999, 2019, Oracle and/or its affiliates. All rights reserved.
3 * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
4 *
5 * This code is free software; you can redistribute it and/or modify it
6 * under the terms of the GNU General Public License version 2 only, as
7 * published by the Free Software Foundation.
8 *
9 * This code is distributed in the hope that it will be useful, but WITHOUT
10 * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
11 * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
12 * version 2 for more details (a copy is included in the LICENSE file that
13 * accompanied this code).
14 *
15 * You should have received a copy of the GNU General Public License version
16 * 2 along with this work; if not, write to the Free Software Foundation,
17 * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
18 *
19 * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
20 * or visit www.oracle.com if you need additional information or have any
21 * questions.
22 *
23 */
24
25 #include "precompiled.hpp"
26 #include "gc/shared/barrierSet.hpp"
27 #include "gc/shared/c2/barrierSetC2.hpp"
28 #include "memory/allocation.inline.hpp"
29 #include "memory/resourceArea.hpp"
30 #include "opto/addnode.hpp"
31 #include "opto/callnode.hpp"
32 #include "opto/castnode.hpp"
33 #include "opto/connode.hpp"
34 #include "opto/castnode.hpp"
35 #include "opto/divnode.hpp"
36 #include "opto/loopnode.hpp"
37 #include "opto/matcher.hpp"
38 #include "opto/mulnode.hpp"
39 #include "opto/movenode.hpp"
40 #include "opto/opaquenode.hpp"
41 #include "opto/rootnode.hpp"
42 #include "opto/subnode.hpp"
43 #include "opto/subtypenode.hpp"
44 #include "utilities/macros.hpp"
45
46 //=============================================================================
47 //------------------------------split_thru_phi---------------------------------
48 // Split Node 'n' through merge point if there is enough win.
49 Node* PhaseIdealLoop::split_thru_phi(Node* n, Node* region, int policy) {
50 if (n->Opcode() == Op_ConvI2L && n->bottom_type() != TypeLong::LONG) {
51 // ConvI2L may have type information on it which is unsafe to push up
52 // so disable this for now
53 return NULL;
54 }
55
56 // Splitting range check CastIIs through a loop induction Phi can
57 // cause new Phis to be created that are left unrelated to the loop
58 // induction Phi and prevent optimizations (vectorization)
59 if (n->Opcode() == Op_CastII && region->is_CountedLoop() &&
60 n->in(1) == region->as_CountedLoop()->phi()) {
61 return NULL;
62 }
63
64 // Bail out if 'n' is a Div or Mod node whose zero check was removed earlier (i.e. control is NULL) and its divisor is an induction variable
65 // phi p of a trip-counted (integer) loop whose inputs could be zero (include zero in their type range). p could have a more precise type
66 // range that does not necessarily include all values of its inputs. Since each of these inputs will be a divisor of the newly cloned nodes
67 // of 'n', we need to bail out of one of these divisors could be zero (zero in its type range).
68 if ((n->Opcode() == Op_DivI || n->Opcode() == Op_ModI) && n->in(0) == NULL
69 && region->is_CountedLoop() && n->in(2) == region->as_CountedLoop()->phi()) {
70 Node* phi = region->as_CountedLoop()->phi();
71 for (uint i = 1; i < phi->req(); i++) {
72 if (_igvn.type(phi->in(i))->filter_speculative(TypeInt::ZERO) != Type::TOP) {
73 // Zero could be a possible value but we already removed the zero check. Bail out to avoid a possible division by zero at a later point.
74 return NULL;
75 }
76 }
77 }
78
79 int wins = 0;
80 assert(!n->is_CFG(), "");
81 assert(region->is_Region(), "");
82
83 const Type* type = n->bottom_type();
84 const TypeOopPtr* t_oop = _igvn.type(n)->isa_oopptr();
85 Node* phi;
86 if (t_oop != NULL && t_oop->is_known_instance_field()) {
87 int iid = t_oop->instance_id();
88 int index = C->get_alias_index(t_oop);
89 int offset = t_oop->offset();
90 phi = new PhiNode(region, type, NULL, iid, index, offset);
91 } else {
92 phi = PhiNode::make_blank(region, n);
93 }
94 uint old_unique = C->unique();
95 for (uint i = 1; i < region->req(); i++) {
96 Node* x;
97 Node* the_clone = NULL;
98 if (region->in(i) == C->top()) {
99 x = C->top(); // Dead path? Use a dead data op
100 } else {
101 x = n->clone(); // Else clone up the data op
102 the_clone = x; // Remember for possible deletion.
103 // Alter data node to use pre-phi inputs
104 if (n->in(0) == region)
105 x->set_req( 0, region->in(i) );
106 for (uint j = 1; j < n->req(); j++) {
107 Node* in = n->in(j);
108 if (in->is_Phi() && in->in(0) == region)
109 x->set_req(j, in->in(i)); // Use pre-Phi input for the clone
110 }
111 }
112 // Check for a 'win' on some paths
113 const Type* t = x->Value(&_igvn);
114
115 bool singleton = t->singleton();
116
117 // A TOP singleton indicates that there are no possible values incoming
118 // along a particular edge. In most cases, this is OK, and the Phi will
119 // be eliminated later in an Ideal call. However, we can't allow this to
120 // happen if the singleton occurs on loop entry, as the elimination of
121 // the PhiNode may cause the resulting node to migrate back to a previous
122 // loop iteration.
123 if (singleton && t == Type::TOP) {
124 // Is_Loop() == false does not confirm the absence of a loop (e.g., an
125 // irreducible loop may not be indicated by an affirmative is_Loop());
126 // therefore, the only top we can split thru a phi is on a backedge of
127 // a loop.
128 singleton &= region->is_Loop() && (i != LoopNode::EntryControl);
129 }
130
131 if (singleton) {
132 wins++;
133 x = ((PhaseGVN&)_igvn).makecon(t);
134 } else {
135 // We now call Identity to try to simplify the cloned node.
136 // Note that some Identity methods call phase->type(this).
137 // Make sure that the type array is big enough for
138 // our new node, even though we may throw the node away.
139 // (Note: This tweaking with igvn only works because x is a new node.)
140 _igvn.set_type(x, t);
141 // If x is a TypeNode, capture any more-precise type permanently into Node
142 // otherwise it will be not updated during igvn->transform since
143 // igvn->type(x) is set to x->Value() already.
144 x->raise_bottom_type(t);
145 Node* y = x->Identity(&_igvn);
146 if (y != x) {
147 wins++;
148 x = y;
149 } else {
150 y = _igvn.hash_find(x);
151 if (y) {
152 wins++;
153 x = y;
154 } else {
155 // Else x is a new node we are keeping
156 // We do not need register_new_node_with_optimizer
157 // because set_type has already been called.
158 _igvn._worklist.push(x);
159 }
160 }
161 }
162 if (x != the_clone && the_clone != NULL)
163 _igvn.remove_dead_node(the_clone);
164 phi->set_req( i, x );
165 }
166 // Too few wins?
167 if (wins <= policy) {
168 _igvn.remove_dead_node(phi);
169 return NULL;
170 }
171
172 // Record Phi
173 register_new_node( phi, region );
174
175 for (uint i2 = 1; i2 < phi->req(); i2++) {
176 Node *x = phi->in(i2);
177 // If we commoned up the cloned 'x' with another existing Node,
178 // the existing Node picks up a new use. We need to make the
179 // existing Node occur higher up so it dominates its uses.
180 Node *old_ctrl;
181 IdealLoopTree *old_loop;
182
183 if (x->is_Con()) {
184 // Constant's control is always root.
185 set_ctrl(x, C->root());
186 continue;
187 }
188 // The occasional new node
189 if (x->_idx >= old_unique) { // Found a new, unplaced node?
190 old_ctrl = NULL;
191 old_loop = NULL; // Not in any prior loop
192 } else {
193 old_ctrl = get_ctrl(x);
194 old_loop = get_loop(old_ctrl); // Get prior loop
195 }
196 // New late point must dominate new use
197 Node *new_ctrl = dom_lca(old_ctrl, region->in(i2));
198 if (new_ctrl == old_ctrl) // Nothing is changed
199 continue;
200
201 IdealLoopTree *new_loop = get_loop(new_ctrl);
202
203 // Don't move x into a loop if its uses are
204 // outside of loop. Otherwise x will be cloned
205 // for each use outside of this loop.
206 IdealLoopTree *use_loop = get_loop(region);
207 if (!new_loop->is_member(use_loop) &&
208 (old_loop == NULL || !new_loop->is_member(old_loop))) {
209 // Take early control, later control will be recalculated
210 // during next iteration of loop optimizations.
211 new_ctrl = get_early_ctrl(x);
212 new_loop = get_loop(new_ctrl);
213 }
214 // Set new location
215 set_ctrl(x, new_ctrl);
216 // If changing loop bodies, see if we need to collect into new body
217 if (old_loop != new_loop) {
218 if (old_loop && !old_loop->_child)
219 old_loop->_body.yank(x);
220 if (!new_loop->_child)
221 new_loop->_body.push(x); // Collect body info
222 }
223 }
224
225 return phi;
226 }
227
228 //------------------------------dominated_by------------------------------------
229 // Replace the dominated test with an obvious true or false. Place it on the
230 // IGVN worklist for later cleanup. Move control-dependent data Nodes on the
231 // live path up to the dominating control.
232 void PhaseIdealLoop::dominated_by( Node *prevdom, Node *iff, bool flip, bool exclude_loop_predicate ) {
233 if (VerifyLoopOptimizations && PrintOpto) { tty->print_cr("dominating test"); }
234
235 // prevdom is the dominating projection of the dominating test.
236 assert(iff->is_If(), "must be");
237 assert(iff->Opcode() == Op_If ||
238 iff->Opcode() == Op_CountedLoopEnd ||
239 iff->Opcode() == Op_LongCountedLoopEnd ||
240 iff->Opcode() == Op_RangeCheck,
241 "Check this code when new subtype is added");
242
243 int pop = prevdom->Opcode();
244 assert( pop == Op_IfFalse || pop == Op_IfTrue, "" );
245 if (flip) {
246 if (pop == Op_IfTrue)
247 pop = Op_IfFalse;
248 else
249 pop = Op_IfTrue;
250 }
251 // 'con' is set to true or false to kill the dominated test.
252 Node *con = _igvn.makecon(pop == Op_IfTrue ? TypeInt::ONE : TypeInt::ZERO);
253 set_ctrl(con, C->root()); // Constant gets a new use
254 // Hack the dominated test
255 _igvn.replace_input_of(iff, 1, con);
256
257 // If I dont have a reachable TRUE and FALSE path following the IfNode then
258 // I can assume this path reaches an infinite loop. In this case it's not
259 // important to optimize the data Nodes - either the whole compilation will
260 // be tossed or this path (and all data Nodes) will go dead.
261 if (iff->outcnt() != 2) return;
262
263 // Make control-dependent data Nodes on the live path (path that will remain
264 // once the dominated IF is removed) become control-dependent on the
265 // dominating projection.
266 Node* dp = iff->as_If()->proj_out_or_null(pop == Op_IfTrue);
267
268 // Loop predicates may have depending checks which should not
269 // be skipped. For example, range check predicate has two checks
270 // for lower and upper bounds.
271 if (dp == NULL)
272 return;
273
274 ProjNode* dp_proj = dp->as_Proj();
275 ProjNode* unc_proj = iff->as_If()->proj_out(1 - dp_proj->_con)->as_Proj();
276 if (exclude_loop_predicate &&
277 (unc_proj->is_uncommon_trap_proj(Deoptimization::Reason_predicate) != NULL ||
278 unc_proj->is_uncommon_trap_proj(Deoptimization::Reason_profile_predicate) != NULL ||
279 unc_proj->is_uncommon_trap_proj(Deoptimization::Reason_range_check) != NULL)) {
280 // If this is a range check (IfNode::is_range_check), do not
281 // reorder because Compile::allow_range_check_smearing might have
282 // changed the check.
283 return; // Let IGVN transformation change control dependence.
284 }
285
286 IdealLoopTree* old_loop = get_loop(dp);
287
288 for (DUIterator_Fast imax, i = dp->fast_outs(imax); i < imax; i++) {
289 Node* cd = dp->fast_out(i); // Control-dependent node
290 // Do not rewire Div and Mod nodes which could have a zero divisor to avoid skipping their zero check.
291 if (cd->depends_only_on_test() && _igvn.no_dependent_zero_check(cd)) {
292 assert(cd->in(0) == dp, "");
293 _igvn.replace_input_of(cd, 0, prevdom);
294 set_early_ctrl(cd, false);
295 IdealLoopTree* new_loop = get_loop(get_ctrl(cd));
296 if (old_loop != new_loop) {
297 if (!old_loop->_child) {
298 old_loop->_body.yank(cd);
299 }
300 if (!new_loop->_child) {
301 new_loop->_body.push(cd);
302 }
303 }
304 --i;
305 --imax;
306 }
307 }
308 }
309
310 //------------------------------has_local_phi_input----------------------------
311 // Return TRUE if 'n' has Phi inputs from its local block and no other
312 // block-local inputs (all non-local-phi inputs come from earlier blocks)
313 Node *PhaseIdealLoop::has_local_phi_input( Node *n ) {
314 Node *n_ctrl = get_ctrl(n);
315 // See if some inputs come from a Phi in this block, or from before
316 // this block.
317 uint i;
318 for( i = 1; i < n->req(); i++ ) {
319 Node *phi = n->in(i);
320 if( phi->is_Phi() && phi->in(0) == n_ctrl )
321 break;
322 }
323 if( i >= n->req() )
324 return NULL; // No Phi inputs; nowhere to clone thru
325
326 // Check for inputs created between 'n' and the Phi input. These
327 // must split as well; they have already been given the chance
328 // (courtesy of a post-order visit) and since they did not we must
329 // recover the 'cost' of splitting them by being very profitable
330 // when splitting 'n'. Since this is unlikely we simply give up.
331 for( i = 1; i < n->req(); i++ ) {
332 Node *m = n->in(i);
333 if( get_ctrl(m) == n_ctrl && !m->is_Phi() ) {
334 // We allow the special case of AddP's with no local inputs.
335 // This allows us to split-up address expressions.
336 if (m->is_AddP() &&
337 get_ctrl(m->in(2)) != n_ctrl &&
338 get_ctrl(m->in(3)) != n_ctrl) {
339 // Move the AddP up to dominating point
340 Node* c = find_non_split_ctrl(idom(n_ctrl));
341 if (c->is_OuterStripMinedLoop()) {
342 c->as_Loop()->verify_strip_mined(1);
343 c = c->in(LoopNode::EntryControl);
344 }
345 set_ctrl_and_loop(m, c);
346 continue;
347 }
348 return NULL;
349 }
350 assert(n->is_Phi() || m->is_Phi() || is_dominator(get_ctrl(m), n_ctrl), "m has strange control");
351 }
352
353 return n_ctrl;
354 }
355
356 //------------------------------remix_address_expressions----------------------
357 // Rework addressing expressions to get the most loop-invariant stuff
358 // moved out. We'd like to do all associative operators, but it's especially
359 // important (common) to do address expressions.
360 Node *PhaseIdealLoop::remix_address_expressions( Node *n ) {
361 if (!has_ctrl(n)) return NULL;
362 Node *n_ctrl = get_ctrl(n);
363 IdealLoopTree *n_loop = get_loop(n_ctrl);
364
365 // See if 'n' mixes loop-varying and loop-invariant inputs and
366 // itself is loop-varying.
367
368 // Only interested in binary ops (and AddP)
369 if( n->req() < 3 || n->req() > 4 ) return NULL;
370
371 Node *n1_ctrl = get_ctrl(n->in( 1));
372 Node *n2_ctrl = get_ctrl(n->in( 2));
373 Node *n3_ctrl = get_ctrl(n->in(n->req() == 3 ? 2 : 3));
374 IdealLoopTree *n1_loop = get_loop( n1_ctrl );
375 IdealLoopTree *n2_loop = get_loop( n2_ctrl );
376 IdealLoopTree *n3_loop = get_loop( n3_ctrl );
377
378 // Does one of my inputs spin in a tighter loop than self?
379 if( (n_loop->is_member( n1_loop ) && n_loop != n1_loop) ||
380 (n_loop->is_member( n2_loop ) && n_loop != n2_loop) ||
381 (n_loop->is_member( n3_loop ) && n_loop != n3_loop) )
382 return NULL; // Leave well enough alone
383
384 // Is at least one of my inputs loop-invariant?
385 if( n1_loop == n_loop &&
386 n2_loop == n_loop &&
387 n3_loop == n_loop )
388 return NULL; // No loop-invariant inputs
389
390
391 int n_op = n->Opcode();
392
393 // Replace expressions like ((V+I) << 2) with (V<<2 + I<<2).
394 if( n_op == Op_LShiftI ) {
395 // Scale is loop invariant
396 Node *scale = n->in(2);
397 Node *scale_ctrl = get_ctrl(scale);
398 IdealLoopTree *scale_loop = get_loop(scale_ctrl );
399 if( n_loop == scale_loop || !scale_loop->is_member( n_loop ) )
400 return NULL;
401 const TypeInt *scale_t = scale->bottom_type()->isa_int();
402 if( scale_t && scale_t->is_con() && scale_t->get_con() >= 16 )
403 return NULL; // Dont bother with byte/short masking
404 // Add must vary with loop (else shift would be loop-invariant)
405 Node *add = n->in(1);
406 Node *add_ctrl = get_ctrl(add);
407 IdealLoopTree *add_loop = get_loop(add_ctrl);
408 //assert( n_loop == add_loop, "" );
409 if( n_loop != add_loop ) return NULL; // happens w/ evil ZKM loops
410
411 // Convert I-V into I+ (0-V); same for V-I
412 if( add->Opcode() == Op_SubI &&
413 _igvn.type( add->in(1) ) != TypeInt::ZERO ) {
414 Node *zero = _igvn.intcon(0);
415 set_ctrl(zero, C->root());
416 Node *neg = new SubINode( _igvn.intcon(0), add->in(2) );
417 register_new_node( neg, get_ctrl(add->in(2) ) );
418 add = new AddINode( add->in(1), neg );
419 register_new_node( add, add_ctrl );
420 }
421 if( add->Opcode() != Op_AddI ) return NULL;
422 // See if one add input is loop invariant
423 Node *add_var = add->in(1);
424 Node *add_var_ctrl = get_ctrl(add_var);
425 IdealLoopTree *add_var_loop = get_loop(add_var_ctrl );
426 Node *add_invar = add->in(2);
427 Node *add_invar_ctrl = get_ctrl(add_invar);
428 IdealLoopTree *add_invar_loop = get_loop(add_invar_ctrl );
429 if( add_var_loop == n_loop ) {
430 } else if( add_invar_loop == n_loop ) {
431 // Swap to find the invariant part
432 add_invar = add_var;
433 add_invar_ctrl = add_var_ctrl;
434 add_invar_loop = add_var_loop;
435 add_var = add->in(2);
436 Node *add_var_ctrl = get_ctrl(add_var);
437 IdealLoopTree *add_var_loop = get_loop(add_var_ctrl );
438 } else // Else neither input is loop invariant
439 return NULL;
440 if( n_loop == add_invar_loop || !add_invar_loop->is_member( n_loop ) )
441 return NULL; // No invariant part of the add?
442
443 // Yes! Reshape address expression!
444 Node *inv_scale = new LShiftINode( add_invar, scale );
445 Node *inv_scale_ctrl =
446 dom_depth(add_invar_ctrl) > dom_depth(scale_ctrl) ?
447 add_invar_ctrl : scale_ctrl;
448 register_new_node( inv_scale, inv_scale_ctrl );
449 Node *var_scale = new LShiftINode( add_var, scale );
450 register_new_node( var_scale, n_ctrl );
451 Node *var_add = new AddINode( var_scale, inv_scale );
452 register_new_node( var_add, n_ctrl );
453 _igvn.replace_node( n, var_add );
454 return var_add;
455 }
456
457 // Replace (I+V) with (V+I)
458 if( n_op == Op_AddI ||
459 n_op == Op_AddL ||
460 n_op == Op_AddF ||
461 n_op == Op_AddD ||
462 n_op == Op_MulI ||
463 n_op == Op_MulL ||
464 n_op == Op_MulF ||
465 n_op == Op_MulD ) {
466 if( n2_loop == n_loop ) {
467 assert( n1_loop != n_loop, "" );
468 n->swap_edges(1, 2);
469 }
470 }
471
472 // Replace ((I1 +p V) +p I2) with ((I1 +p I2) +p V),
473 // but not if I2 is a constant.
474 if( n_op == Op_AddP ) {
475 if( n2_loop == n_loop && n3_loop != n_loop ) {
476 if( n->in(2)->Opcode() == Op_AddP && !n->in(3)->is_Con() ) {
477 Node *n22_ctrl = get_ctrl(n->in(2)->in(2));
478 Node *n23_ctrl = get_ctrl(n->in(2)->in(3));
479 IdealLoopTree *n22loop = get_loop( n22_ctrl );
480 IdealLoopTree *n23_loop = get_loop( n23_ctrl );
481 if( n22loop != n_loop && n22loop->is_member(n_loop) &&
482 n23_loop == n_loop ) {
483 Node *add1 = new AddPNode( n->in(1), n->in(2)->in(2), n->in(3) );
484 // Stuff new AddP in the loop preheader
485 register_new_node( add1, n_loop->_head->in(LoopNode::EntryControl) );
486 Node *add2 = new AddPNode( n->in(1), add1, n->in(2)->in(3) );
487 register_new_node( add2, n_ctrl );
488 _igvn.replace_node( n, add2 );
489 return add2;
490 }
491 }
492 }
493
494 // Replace (I1 +p (I2 + V)) with ((I1 +p I2) +p V)
495 if (n2_loop != n_loop && n3_loop == n_loop) {
496 if (n->in(3)->Opcode() == Op_AddX) {
497 Node *V = n->in(3)->in(1);
498 Node *I = n->in(3)->in(2);
499 if (is_member(n_loop,get_ctrl(V))) {
500 } else {
501 Node *tmp = V; V = I; I = tmp;
502 }
503 if (!is_member(n_loop,get_ctrl(I))) {
504 Node *add1 = new AddPNode(n->in(1), n->in(2), I);
505 // Stuff new AddP in the loop preheader
506 register_new_node(add1, n_loop->_head->in(LoopNode::EntryControl));
507 Node *add2 = new AddPNode(n->in(1), add1, V);
508 register_new_node(add2, n_ctrl);
509 _igvn.replace_node(n, add2);
510 return add2;
511 }
512 }
513 }
514 }
515
516 return NULL;
517 }
518
519 // Optimize ((in1[2*i] * in2[2*i]) + (in1[2*i+1] * in2[2*i+1]))
520 Node *PhaseIdealLoop::convert_add_to_muladd(Node* n) {
521 assert(n->Opcode() == Op_AddI, "sanity");
522 Node * nn = NULL;
523 Node * in1 = n->in(1);
524 Node * in2 = n->in(2);
525 if (in1->Opcode() == Op_MulI && in2->Opcode() == Op_MulI) {
526 IdealLoopTree* loop_n = get_loop(get_ctrl(n));
527 if (loop_n->is_counted() &&
528 loop_n->_head->as_Loop()->is_valid_counted_loop(T_INT) &&
529 Matcher::match_rule_supported(Op_MulAddVS2VI) &&
530 Matcher::match_rule_supported(Op_MulAddS2I)) {
531 Node* mul_in1 = in1->in(1);
532 Node* mul_in2 = in1->in(2);
533 Node* mul_in3 = in2->in(1);
534 Node* mul_in4 = in2->in(2);
535 if (mul_in1->Opcode() == Op_LoadS &&
536 mul_in2->Opcode() == Op_LoadS &&
537 mul_in3->Opcode() == Op_LoadS &&
538 mul_in4->Opcode() == Op_LoadS) {
539 IdealLoopTree* loop1 = get_loop(get_ctrl(mul_in1));
540 IdealLoopTree* loop2 = get_loop(get_ctrl(mul_in2));
541 IdealLoopTree* loop3 = get_loop(get_ctrl(mul_in3));
542 IdealLoopTree* loop4 = get_loop(get_ctrl(mul_in4));
543 IdealLoopTree* loop5 = get_loop(get_ctrl(in1));
544 IdealLoopTree* loop6 = get_loop(get_ctrl(in2));
545 // All nodes should be in the same counted loop.
546 if (loop_n == loop1 && loop_n == loop2 && loop_n == loop3 &&
547 loop_n == loop4 && loop_n == loop5 && loop_n == loop6) {
548 Node* adr1 = mul_in1->in(MemNode::Address);
549 Node* adr2 = mul_in2->in(MemNode::Address);
550 Node* adr3 = mul_in3->in(MemNode::Address);
551 Node* adr4 = mul_in4->in(MemNode::Address);
552 if (adr1->is_AddP() && adr2->is_AddP() && adr3->is_AddP() && adr4->is_AddP()) {
553 if ((adr1->in(AddPNode::Base) == adr3->in(AddPNode::Base)) &&
554 (adr2->in(AddPNode::Base) == adr4->in(AddPNode::Base))) {
555 nn = new MulAddS2INode(mul_in1, mul_in2, mul_in3, mul_in4);
556 register_new_node(nn, get_ctrl(n));
557 _igvn.replace_node(n, nn);
558 return nn;
559 } else if ((adr1->in(AddPNode::Base) == adr4->in(AddPNode::Base)) &&
560 (adr2->in(AddPNode::Base) == adr3->in(AddPNode::Base))) {
561 nn = new MulAddS2INode(mul_in1, mul_in2, mul_in4, mul_in3);
562 register_new_node(nn, get_ctrl(n));
563 _igvn.replace_node(n, nn);
564 return nn;
565 }
566 }
567 }
568 }
569 }
570 }
571 return nn;
572 }
573
574 //------------------------------conditional_move-------------------------------
575 // Attempt to replace a Phi with a conditional move. We have some pretty
576 // strict profitability requirements. All Phis at the merge point must
577 // be converted, so we can remove the control flow. We need to limit the
578 // number of c-moves to a small handful. All code that was in the side-arms
579 // of the CFG diamond is now speculatively executed. This code has to be
580 // "cheap enough". We are pretty much limited to CFG diamonds that merge
581 // 1 or 2 items with a total of 1 or 2 ops executed speculatively.
582 Node *PhaseIdealLoop::conditional_move( Node *region ) {
583
584 assert(region->is_Region(), "sanity check");
585 if (region->req() != 3) return NULL;
586
587 // Check for CFG diamond
588 Node *lp = region->in(1);
589 Node *rp = region->in(2);
590 if (!lp || !rp) return NULL;
591 Node *lp_c = lp->in(0);
592 if (lp_c == NULL || lp_c != rp->in(0) || !lp_c->is_If()) return NULL;
593 IfNode *iff = lp_c->as_If();
594
595 // Check for ops pinned in an arm of the diamond.
596 // Can't remove the control flow in this case
597 if (lp->outcnt() > 1) return NULL;
598 if (rp->outcnt() > 1) return NULL;
599
600 IdealLoopTree* r_loop = get_loop(region);
601 assert(r_loop == get_loop(iff), "sanity");
602 // Always convert to CMOVE if all results are used only outside this loop.
603 bool used_inside_loop = (r_loop == _ltree_root);
604
605 // Check profitability
606 int cost = 0;
607 int phis = 0;
608 for (DUIterator_Fast imax, i = region->fast_outs(imax); i < imax; i++) {
609 Node *out = region->fast_out(i);
610 if (!out->is_Phi()) continue; // Ignore other control edges, etc
611 phis++;
612 PhiNode* phi = out->as_Phi();
613 BasicType bt = phi->type()->basic_type();
614 switch (bt) {
615 case T_DOUBLE:
616 case T_FLOAT:
617 if (C->use_cmove()) {
618 continue; //TODO: maybe we want to add some cost
619 }
620 cost += Matcher::float_cmove_cost(); // Could be very expensive
621 break;
622 case T_LONG: {
623 cost += Matcher::long_cmove_cost(); // May encodes as 2 CMOV's
624 }
625 case T_INT: // These all CMOV fine
626 case T_ADDRESS: { // (RawPtr)
627 cost++;
628 break;
629 }
630 case T_NARROWOOP: // Fall through
631 case T_OBJECT: { // Base oops are OK, but not derived oops
632 const TypeOopPtr *tp = phi->type()->make_ptr()->isa_oopptr();
633 // Derived pointers are Bad (tm): what's the Base (for GC purposes) of a
634 // CMOVE'd derived pointer? It's a CMOVE'd derived base. Thus
635 // CMOVE'ing a derived pointer requires we also CMOVE the base. If we
636 // have a Phi for the base here that we convert to a CMOVE all is well
637 // and good. But if the base is dead, we'll not make a CMOVE. Later
638 // the allocator will have to produce a base by creating a CMOVE of the
639 // relevant bases. This puts the allocator in the business of
640 // manufacturing expensive instructions, generally a bad plan.
641 // Just Say No to Conditionally-Moved Derived Pointers.
642 if (tp && tp->offset() != 0)
643 return NULL;
644 cost++;
645 break;
646 }
647 default:
648 return NULL; // In particular, can't do memory or I/O
649 }
650 // Add in cost any speculative ops
651 for (uint j = 1; j < region->req(); j++) {
652 Node *proj = region->in(j);
653 Node *inp = phi->in(j);
654 if (get_ctrl(inp) == proj) { // Found local op
655 cost++;
656 // Check for a chain of dependent ops; these will all become
657 // speculative in a CMOV.
658 for (uint k = 1; k < inp->req(); k++)
659 if (get_ctrl(inp->in(k)) == proj)
660 cost += ConditionalMoveLimit; // Too much speculative goo
661 }
662 }
663 // See if the Phi is used by a Cmp or Narrow oop Decode/Encode.
664 // This will likely Split-If, a higher-payoff operation.
665 for (DUIterator_Fast kmax, k = phi->fast_outs(kmax); k < kmax; k++) {
666 Node* use = phi->fast_out(k);
667 if (use->is_Cmp() || use->is_DecodeNarrowPtr() || use->is_EncodeNarrowPtr())
668 cost += ConditionalMoveLimit;
669 // Is there a use inside the loop?
670 // Note: check only basic types since CMoveP is pinned.
671 if (!used_inside_loop && is_java_primitive(bt)) {
672 IdealLoopTree* u_loop = get_loop(has_ctrl(use) ? get_ctrl(use) : use);
673 if (r_loop == u_loop || r_loop->is_member(u_loop)) {
674 used_inside_loop = true;
675 }
676 }
677 }
678 }//for
679 Node* bol = iff->in(1);
680 if (bol->Opcode() == Op_Opaque4) {
681 return NULL; // Ignore loop predicate checks (the Opaque4 ensures they will go away)
682 }
683 assert(bol->Opcode() == Op_Bool, "Unexpected node");
684 int cmp_op = bol->in(1)->Opcode();
685 if (cmp_op == Op_SubTypeCheck) { // SubTypeCheck expansion expects an IfNode
686 return NULL;
687 }
688 // It is expensive to generate flags from a float compare.
689 // Avoid duplicated float compare.
690 if (phis > 1 && (cmp_op == Op_CmpF || cmp_op == Op_CmpD)) return NULL;
691
692 float infrequent_prob = PROB_UNLIKELY_MAG(3);
693 // Ignore cost and blocks frequency if CMOVE can be moved outside the loop.
694 if (used_inside_loop) {
695 if (cost >= ConditionalMoveLimit) return NULL; // Too much goo
696
697 // BlockLayoutByFrequency optimization moves infrequent branch
698 // from hot path. No point in CMOV'ing in such case (110 is used
699 // instead of 100 to take into account not exactness of float value).
700 if (BlockLayoutByFrequency) {
701 infrequent_prob = MAX2(infrequent_prob, (float)BlockLayoutMinDiamondPercentage/110.0f);
702 }
703 }
704 // Check for highly predictable branch. No point in CMOV'ing if
705 // we are going to predict accurately all the time.
706 if (C->use_cmove() && (cmp_op == Op_CmpF || cmp_op == Op_CmpD)) {
707 //keep going
708 } else if (iff->_prob < infrequent_prob ||
709 iff->_prob > (1.0f - infrequent_prob))
710 return NULL;
711
712 // --------------
713 // Now replace all Phis with CMOV's
714 Node *cmov_ctrl = iff->in(0);
715 uint flip = (lp->Opcode() == Op_IfTrue);
716 Node_List wq;
717 while (1) {
718 PhiNode* phi = NULL;
719 for (DUIterator_Fast imax, i = region->fast_outs(imax); i < imax; i++) {
720 Node *out = region->fast_out(i);
721 if (out->is_Phi()) {
722 phi = out->as_Phi();
723 break;
724 }
725 }
726 if (phi == NULL || _igvn.type(phi) == Type::TOP) {
727 break;
728 }
729 if (PrintOpto && VerifyLoopOptimizations) { tty->print_cr("CMOV"); }
730 // Move speculative ops
731 wq.push(phi);
732 while (wq.size() > 0) {
733 Node *n = wq.pop();
734 for (uint j = 1; j < n->req(); j++) {
735 Node* m = n->in(j);
736 if (m != NULL && !is_dominator(get_ctrl(m), cmov_ctrl)) {
737 #ifndef PRODUCT
738 if (PrintOpto && VerifyLoopOptimizations) {
739 tty->print(" speculate: ");
740 m->dump();
741 }
742 #endif
743 set_ctrl(m, cmov_ctrl);
744 wq.push(m);
745 }
746 }
747 }
748 Node *cmov = CMoveNode::make(cmov_ctrl, iff->in(1), phi->in(1+flip), phi->in(2-flip), _igvn.type(phi));
749 register_new_node( cmov, cmov_ctrl );
750 _igvn.replace_node( phi, cmov );
751 #ifndef PRODUCT
752 if (TraceLoopOpts) {
753 tty->print("CMOV ");
754 r_loop->dump_head();
755 if (Verbose) {
756 bol->in(1)->dump(1);
757 cmov->dump(1);
758 }
759 }
760 if (VerifyLoopOptimizations) verify();
761 #endif
762 }
763
764 // The useless CFG diamond will fold up later; see the optimization in
765 // RegionNode::Ideal.
766 _igvn._worklist.push(region);
767
768 return iff->in(1);
769 }
770
771 static void enqueue_cfg_uses(Node* m, Unique_Node_List& wq) {
772 for (DUIterator_Fast imax, i = m->fast_outs(imax); i < imax; i++) {
773 Node* u = m->fast_out(i);
774 if (u->is_CFG()) {
775 if (u->Opcode() == Op_NeverBranch) {
776 u = ((NeverBranchNode*)u)->proj_out(0);
777 enqueue_cfg_uses(u, wq);
778 } else {
779 wq.push(u);
780 }
781 }
782 }
783 }
784
785 // Try moving a store out of a loop, right before the loop
786 Node* PhaseIdealLoop::try_move_store_before_loop(Node* n, Node *n_ctrl) {
787 // Store has to be first in the loop body
788 IdealLoopTree *n_loop = get_loop(n_ctrl);
789 if (n->is_Store() && n_loop != _ltree_root &&
790 n_loop->is_loop() && n_loop->_head->is_Loop() &&
791 n->in(0) != NULL) {
792 Node* address = n->in(MemNode::Address);
793 Node* value = n->in(MemNode::ValueIn);
794 Node* mem = n->in(MemNode::Memory);
795 IdealLoopTree* address_loop = get_loop(get_ctrl(address));
796 IdealLoopTree* value_loop = get_loop(get_ctrl(value));
797
798 // - address and value must be loop invariant
799 // - memory must be a memory Phi for the loop
800 // - Store must be the only store on this memory slice in the
801 // loop: if there's another store following this one then value
802 // written at iteration i by the second store could be overwritten
803 // at iteration i+n by the first store: it's not safe to move the
804 // first store out of the loop
805 // - nothing must observe the memory Phi: it guarantees no read
806 // before the store, we are also guaranteed the store post
807 // dominates the loop head (ignoring a possible early
808 // exit). Otherwise there would be extra Phi involved between the
809 // loop's Phi and the store.
810 // - there must be no early exit from the loop before the Store
811 // (such an exit most of the time would be an extra use of the
812 // memory Phi but sometimes is a bottom memory Phi that takes the
813 // store as input).
814
815 if (!n_loop->is_member(address_loop) &&
816 !n_loop->is_member(value_loop) &&
817 mem->is_Phi() && mem->in(0) == n_loop->_head &&
818 mem->outcnt() == 1 &&
819 mem->in(LoopNode::LoopBackControl) == n) {
820
821 assert(n_loop->_tail != NULL, "need a tail");
822 assert(is_dominator(n_ctrl, n_loop->_tail), "store control must not be in a branch in the loop");
823
824 // Verify that there's no early exit of the loop before the store.
825 bool ctrl_ok = false;
826 {
827 // Follow control from loop head until n, we exit the loop or
828 // we reach the tail
829 ResourceMark rm;
830 Unique_Node_List wq;
831 wq.push(n_loop->_head);
832
833 for (uint next = 0; next < wq.size(); ++next) {
834 Node *m = wq.at(next);
835 if (m == n->in(0)) {
836 ctrl_ok = true;
837 continue;
838 }
839 assert(!has_ctrl(m), "should be CFG");
840 if (!n_loop->is_member(get_loop(m)) || m == n_loop->_tail) {
841 ctrl_ok = false;
842 break;
843 }
844 enqueue_cfg_uses(m, wq);
845 if (wq.size() > 10) {
846 ctrl_ok = false;
847 break;
848 }
849 }
850 }
851 if (ctrl_ok) {
852 // move the Store
853 _igvn.replace_input_of(mem, LoopNode::LoopBackControl, mem);
854 _igvn.replace_input_of(n, 0, n_loop->_head->as_Loop()->skip_strip_mined()->in(LoopNode::EntryControl));
855 _igvn.replace_input_of(n, MemNode::Memory, mem->in(LoopNode::EntryControl));
856 // Disconnect the phi now. An empty phi can confuse other
857 // optimizations in this pass of loop opts.
858 _igvn.replace_node(mem, mem->in(LoopNode::EntryControl));
859 n_loop->_body.yank(mem);
860
861 set_ctrl_and_loop(n, n->in(0));
862
863 return n;
864 }
865 }
866 }
867 return NULL;
868 }
869
870 // Try moving a store out of a loop, right after the loop
871 void PhaseIdealLoop::try_move_store_after_loop(Node* n) {
872 if (n->is_Store() && n->in(0) != NULL) {
873 Node *n_ctrl = get_ctrl(n);
874 IdealLoopTree *n_loop = get_loop(n_ctrl);
875 // Store must be in a loop
876 if (n_loop != _ltree_root && !n_loop->_irreducible) {
877 Node* address = n->in(MemNode::Address);
878 Node* value = n->in(MemNode::ValueIn);
879 IdealLoopTree* address_loop = get_loop(get_ctrl(address));
880 // address must be loop invariant
881 if (!n_loop->is_member(address_loop)) {
882 // Store must be last on this memory slice in the loop and
883 // nothing in the loop must observe it
884 Node* phi = NULL;
885 for (DUIterator_Fast imax, i = n->fast_outs(imax); i < imax; i++) {
886 Node* u = n->fast_out(i);
887 if (has_ctrl(u)) { // control use?
888 IdealLoopTree *u_loop = get_loop(get_ctrl(u));
889 if (!n_loop->is_member(u_loop)) {
890 continue;
891 }
892 if (u->is_Phi() && u->in(0) == n_loop->_head) {
893 assert(_igvn.type(u) == Type::MEMORY, "bad phi");
894 // multiple phis on the same slice are possible
895 if (phi != NULL) {
896 return;
897 }
898 phi = u;
899 continue;
900 }
901 }
902 return;
903 }
904 if (phi != NULL) {
905 // Nothing in the loop before the store (next iteration)
906 // must observe the stored value
907 bool mem_ok = true;
908 {
909 ResourceMark rm;
910 Unique_Node_List wq;
911 wq.push(phi);
912 for (uint next = 0; next < wq.size() && mem_ok; ++next) {
913 Node *m = wq.at(next);
914 for (DUIterator_Fast imax, i = m->fast_outs(imax); i < imax && mem_ok; i++) {
915 Node* u = m->fast_out(i);
916 if (u->is_Store() || u->is_Phi()) {
917 if (u != n) {
918 wq.push(u);
919 mem_ok = (wq.size() <= 10);
920 }
921 } else {
922 mem_ok = false;
923 break;
924 }
925 }
926 }
927 }
928 if (mem_ok) {
929 // Move the store out of the loop if the LCA of all
930 // users (except for the phi) is outside the loop.
931 Node* hook = new Node(1);
932 hook->init_req(0, n_ctrl); // Add an input to prevent hook from being dead
933 _igvn.rehash_node_delayed(phi);
934 int count = phi->replace_edge(n, hook, &_igvn);
935 assert(count > 0, "inconsistent phi");
936
937 // Compute latest point this store can go
938 Node* lca = get_late_ctrl(n, get_ctrl(n));
939 if (lca->is_OuterStripMinedLoop()) {
940 lca = lca->in(LoopNode::EntryControl);
941 }
942 if (n_loop->is_member(get_loop(lca))) {
943 // LCA is in the loop - bail out
944 _igvn.replace_node(hook, n);
945 return;
946 }
947 #ifdef ASSERT
948 if (n_loop->_head->is_Loop() && n_loop->_head->as_Loop()->is_strip_mined()) {
949 assert(n_loop->_head->Opcode() == Op_CountedLoop, "outer loop is a strip mined");
950 n_loop->_head->as_Loop()->verify_strip_mined(1);
951 Node* outer = n_loop->_head->as_CountedLoop()->outer_loop();
952 IdealLoopTree* outer_loop = get_loop(outer);
953 assert(n_loop->_parent == outer_loop, "broken loop tree");
954 assert(get_loop(lca) == outer_loop, "safepoint in outer loop consume all memory state");
955 }
956 #endif
957 lca = place_outside_loop(lca, n_loop);
958 assert(!n_loop->is_member(get_loop(lca)), "control must not be back in the loop");
959 assert(get_loop(lca)->_nest < n_loop->_nest || lca->in(0)->Opcode() == Op_NeverBranch, "must not be moved into inner loop");
960
961 // Move store out of the loop
962 _igvn.replace_node(hook, n->in(MemNode::Memory));
963 _igvn.replace_input_of(n, 0, lca);
964 set_ctrl_and_loop(n, lca);
965
966 // Disconnect the phi now. An empty phi can confuse other
967 // optimizations in this pass of loop opts..
968 if (phi->in(LoopNode::LoopBackControl) == phi) {
969 _igvn.replace_node(phi, phi->in(LoopNode::EntryControl));
970 n_loop->_body.yank(phi);
971 }
972 }
973 }
974 }
975 }
976 }
977 }
978
979 //------------------------------split_if_with_blocks_pre-----------------------
980 // Do the real work in a non-recursive function. Data nodes want to be
981 // cloned in the pre-order so they can feed each other nicely.
982 Node *PhaseIdealLoop::split_if_with_blocks_pre( Node *n ) {
983 // Cloning these guys is unlikely to win
984 int n_op = n->Opcode();
985 if (n_op == Op_MergeMem) {
986 return n;
987 }
988 if (n->is_Proj()) {
989 return n;
990 }
991 // Do not clone-up CmpFXXX variations, as these are always
992 // followed by a CmpI
993 if (n->is_Cmp()) {
994 return n;
995 }
996 // Attempt to use a conditional move instead of a phi/branch
997 if (ConditionalMoveLimit > 0 && n_op == Op_Region) {
998 Node *cmov = conditional_move( n );
999 if (cmov) {
1000 return cmov;
1001 }
1002 }
1003 if (n->is_CFG() || n->is_LoadStore()) {
1004 return n;
1005 }
1006 if (n->is_Opaque1() || // Opaque nodes cannot be mod'd
1007 n_op == Op_Opaque2) {
1008 if (!C->major_progress()) { // If chance of no more loop opts...
1009 _igvn._worklist.push(n); // maybe we'll remove them
1010 }
1011 return n;
1012 }
1013
1014 if (n->is_Con()) {
1015 return n; // No cloning for Con nodes
1016 }
1017
1018 Node *n_ctrl = get_ctrl(n);
1019 if (!n_ctrl) {
1020 return n; // Dead node
1021 }
1022
1023 Node* res = try_move_store_before_loop(n, n_ctrl);
1024 if (res != NULL) {
1025 return n;
1026 }
1027
1028 // Attempt to remix address expressions for loop invariants
1029 Node *m = remix_address_expressions( n );
1030 if( m ) return m;
1031
1032 if (n_op == Op_AddI) {
1033 Node *nn = convert_add_to_muladd( n );
1034 if ( nn ) return nn;
1035 }
1036
1037 if (n->is_ConstraintCast()) {
1038 Node* dom_cast = n->as_ConstraintCast()->dominating_cast(&_igvn, this);
1039 // ConstraintCastNode::dominating_cast() uses node control input to determine domination.
1040 // Node control inputs don't necessarily agree with loop control info (due to
1041 // transformations happened in between), thus additional dominance check is needed
1042 // to keep loop info valid.
1043 if (dom_cast != NULL && is_dominator(get_ctrl(dom_cast), get_ctrl(n))) {
1044 _igvn.replace_node(n, dom_cast);
1045 return dom_cast;
1046 }
1047 }
1048
1049 // Determine if the Node has inputs from some local Phi.
1050 // Returns the block to clone thru.
1051 Node *n_blk = has_local_phi_input( n );
1052 if( !n_blk ) return n;
1053
1054 // Do not clone the trip counter through on a CountedLoop
1055 // (messes up the canonical shape).
1056 if (((n_blk->is_CountedLoop() || (n_blk->is_Loop() && n_blk->as_Loop()->is_transformed_long_inner_loop())) && n->Opcode() == Op_AddI) ||
1057 (n_blk->is_LongCountedLoop() && n->Opcode() == Op_AddL)) {
1058 return n;
1059 }
1060
1061 // Check for having no control input; not pinned. Allow
1062 // dominating control.
1063 if (n->in(0)) {
1064 Node *dom = idom(n_blk);
1065 if (dom_lca(n->in(0), dom) != n->in(0)) {
1066 return n;
1067 }
1068 }
1069 // Policy: when is it profitable. You must get more wins than
1070 // policy before it is considered profitable. Policy is usually 0,
1071 // so 1 win is considered profitable. Big merges will require big
1072 // cloning, so get a larger policy.
1073 int policy = n_blk->req() >> 2;
1074
1075 // If the loop is a candidate for range check elimination,
1076 // delay splitting through it's phi until a later loop optimization
1077 if (n_blk->is_CountedLoop()) {
1078 IdealLoopTree *lp = get_loop(n_blk);
1079 if (lp && lp->_rce_candidate) {
1080 return n;
1081 }
1082 }
1083
1084 if (must_throttle_split_if()) return n;
1085
1086 // Split 'n' through the merge point if it is profitable
1087 Node *phi = split_thru_phi( n, n_blk, policy );
1088 if (!phi) return n;
1089
1090 // Found a Phi to split thru!
1091 // Replace 'n' with the new phi
1092 _igvn.replace_node( n, phi );
1093 // Moved a load around the loop, 'en-registering' something.
1094 if (n_blk->is_Loop() && n->is_Load() &&
1095 !phi->in(LoopNode::LoopBackControl)->is_Load())
1096 C->set_major_progress();
1097
1098 return phi;
1099 }
1100
1101 static bool merge_point_too_heavy(Compile* C, Node* region) {
1102 // Bail out if the region and its phis have too many users.
1103 int weight = 0;
1104 for (DUIterator_Fast imax, i = region->fast_outs(imax); i < imax; i++) {
1105 weight += region->fast_out(i)->outcnt();
1106 }
1107 int nodes_left = C->max_node_limit() - C->live_nodes();
1108 if (weight * 8 > nodes_left) {
1109 if (PrintOpto) {
1110 tty->print_cr("*** Split-if bails out: %d nodes, region weight %d", C->unique(), weight);
1111 }
1112 return true;
1113 } else {
1114 return false;
1115 }
1116 }
1117
1118 static bool merge_point_safe(Node* region) {
1119 // 4799512: Stop split_if_with_blocks from splitting a block with a ConvI2LNode
1120 // having a PhiNode input. This sidesteps the dangerous case where the split
1121 // ConvI2LNode may become TOP if the input Value() does not
1122 // overlap the ConvI2L range, leaving a node which may not dominate its
1123 // uses.
1124 // A better fix for this problem can be found in the BugTraq entry, but
1125 // expediency for Mantis demands this hack.
1126 #ifdef _LP64
1127 for (DUIterator_Fast imax, i = region->fast_outs(imax); i < imax; i++) {
1128 Node* n = region->fast_out(i);
1129 if (n->is_Phi()) {
1130 for (DUIterator_Fast jmax, j = n->fast_outs(jmax); j < jmax; j++) {
1131 Node* m = n->fast_out(j);
1132 if (m->Opcode() == Op_ConvI2L)
1133 return false;
1134 if (m->is_CastII()) {
1135 return false;
1136 }
1137 }
1138 }
1139 }
1140 #endif
1141 return true;
1142 }
1143
1144
1145 //------------------------------place_outside_loop---------------------------------
1146 // Place some computation outside of this loop on the path to the use passed as argument
1147 Node* PhaseIdealLoop::place_outside_loop(Node* useblock, IdealLoopTree* loop) const {
1148 Node* head = loop->_head;
1149 assert(!loop->is_member(get_loop(useblock)), "must be outside loop");
1150 if (head->is_Loop() && head->as_Loop()->is_strip_mined()) {
1151 loop = loop->_parent;
1152 assert(loop->_head->is_OuterStripMinedLoop(), "malformed strip mined loop");
1153 }
1154
1155 // Pick control right outside the loop
1156 for (;;) {
1157 Node* dom = idom(useblock);
1158 if (loop->is_member(get_loop(dom)) ||
1159 // NeverBranch nodes are not assigned to the loop when constructed
1160 (dom->Opcode() == Op_NeverBranch && loop->is_member(get_loop(dom->in(0))))) {
1161 break;
1162 }
1163 useblock = dom;
1164 }
1165 assert(find_non_split_ctrl(useblock) == useblock, "should be non split control");
1166 return useblock;
1167 }
1168
1169
1170 bool PhaseIdealLoop::identical_backtoback_ifs(Node *n) {
1171 if (!n->is_If() || n->is_BaseCountedLoopEnd()) {
1172 return false;
1173 }
1174 if (!n->in(0)->is_Region()) {
1175 return false;
1176 }
1177 Node* region = n->in(0);
1178 Node* dom = idom(region);
1179 if (!dom->is_If() || dom->in(1) != n->in(1)) {
1180 return false;
1181 }
1182 IfNode* dom_if = dom->as_If();
1183 Node* proj_true = dom_if->proj_out(1);
1184 Node* proj_false = dom_if->proj_out(0);
1185
1186 for (uint i = 1; i < region->req(); i++) {
1187 if (is_dominator(proj_true, region->in(i))) {
1188 continue;
1189 }
1190 if (is_dominator(proj_false, region->in(i))) {
1191 continue;
1192 }
1193 return false;
1194 }
1195
1196 return true;
1197 }
1198
1199
1200 bool PhaseIdealLoop::can_split_if(Node* n_ctrl) {
1201 if (must_throttle_split_if()) {
1202 return false;
1203 }
1204
1205 // Do not do 'split-if' if irreducible loops are present.
1206 if (_has_irreducible_loops) {
1207 return false;
1208 }
1209
1210 if (merge_point_too_heavy(C, n_ctrl)) {
1211 return false;
1212 }
1213
1214 // Do not do 'split-if' if some paths are dead. First do dead code
1215 // elimination and then see if its still profitable.
1216 for (uint i = 1; i < n_ctrl->req(); i++) {
1217 if (n_ctrl->in(i) == C->top()) {
1218 return false;
1219 }
1220 }
1221
1222 // If trying to do a 'Split-If' at the loop head, it is only
1223 // profitable if the cmp folds up on BOTH paths. Otherwise we
1224 // risk peeling a loop forever.
1225
1226 // CNC - Disabled for now. Requires careful handling of loop
1227 // body selection for the cloned code. Also, make sure we check
1228 // for any input path not being in the same loop as n_ctrl. For
1229 // irreducible loops we cannot check for 'n_ctrl->is_Loop()'
1230 // because the alternative loop entry points won't be converted
1231 // into LoopNodes.
1232 IdealLoopTree *n_loop = get_loop(n_ctrl);
1233 for (uint j = 1; j < n_ctrl->req(); j++) {
1234 if (get_loop(n_ctrl->in(j)) != n_loop) {
1235 return false;
1236 }
1237 }
1238
1239 // Check for safety of the merge point.
1240 if (!merge_point_safe(n_ctrl)) {
1241 return false;
1242 }
1243
1244 return true;
1245 }
1246
1247 // Detect if the node is the inner strip-mined loop
1248 // Return: NULL if it's not the case, or the exit of outer strip-mined loop
1249 static Node* is_inner_of_stripmined_loop(const Node* out) {
1250 Node* out_le = NULL;
1251
1252 if (out->is_CountedLoopEnd()) {
1253 const CountedLoopNode* loop = out->as_CountedLoopEnd()->loopnode();
1254
1255 if (loop != NULL && loop->is_strip_mined()) {
1256 out_le = loop->in(LoopNode::EntryControl)->as_OuterStripMinedLoop()->outer_loop_exit();
1257 }
1258 }
1259
1260 return out_le;
1261 }
1262
1263 //------------------------------split_if_with_blocks_post----------------------
1264 // Do the real work in a non-recursive function. CFG hackery wants to be
1265 // in the post-order, so it can dirty the I-DOM info and not use the dirtied
1266 // info.
1267 void PhaseIdealLoop::split_if_with_blocks_post(Node *n) {
1268
1269 // Cloning Cmp through Phi's involves the split-if transform.
1270 // FastLock is not used by an If
1271 if (n->is_Cmp() && !n->is_FastLock()) {
1272 Node *n_ctrl = get_ctrl(n);
1273 // Determine if the Node has inputs from some local Phi.
1274 // Returns the block to clone thru.
1275 Node *n_blk = has_local_phi_input(n);
1276 if (n_blk != n_ctrl) {
1277 return;
1278 }
1279
1280 if (!can_split_if(n_ctrl)) {
1281 return;
1282 }
1283
1284 if (n->outcnt() != 1) {
1285 return; // Multiple bool's from 1 compare?
1286 }
1287 Node *bol = n->unique_out();
1288 assert(bol->is_Bool(), "expect a bool here");
1289 if (bol->outcnt() != 1) {
1290 return;// Multiple branches from 1 compare?
1291 }
1292 Node *iff = bol->unique_out();
1293
1294 // Check some safety conditions
1295 if (iff->is_If()) { // Classic split-if?
1296 if (iff->in(0) != n_ctrl) {
1297 return; // Compare must be in same blk as if
1298 }
1299 } else if (iff->is_CMove()) { // Trying to split-up a CMOVE
1300 // Can't split CMove with different control edge.
1301 if (iff->in(0) != NULL && iff->in(0) != n_ctrl ) {
1302 return;
1303 }
1304 if (get_ctrl(iff->in(2)) == n_ctrl ||
1305 get_ctrl(iff->in(3)) == n_ctrl) {
1306 return; // Inputs not yet split-up
1307 }
1308 if (get_loop(n_ctrl) != get_loop(get_ctrl(iff))) {
1309 return; // Loop-invar test gates loop-varying CMOVE
1310 }
1311 } else {
1312 return; // some other kind of node, such as an Allocate
1313 }
1314
1315 // When is split-if profitable? Every 'win' on means some control flow
1316 // goes dead, so it's almost always a win.
1317 int policy = 0;
1318 // Split compare 'n' through the merge point if it is profitable
1319 Node *phi = split_thru_phi( n, n_ctrl, policy);
1320 if (!phi) {
1321 return;
1322 }
1323
1324 // Found a Phi to split thru!
1325 // Replace 'n' with the new phi
1326 _igvn.replace_node(n, phi);
1327
1328 // Now split the bool up thru the phi
1329 Node *bolphi = split_thru_phi(bol, n_ctrl, -1);
1330 guarantee(bolphi != NULL, "null boolean phi node");
1331
1332 _igvn.replace_node(bol, bolphi);
1333 assert(iff->in(1) == bolphi, "");
1334
1335 if (bolphi->Value(&_igvn)->singleton()) {
1336 return;
1337 }
1338
1339 // Conditional-move? Must split up now
1340 if (!iff->is_If()) {
1341 Node *cmovphi = split_thru_phi(iff, n_ctrl, -1);
1342 _igvn.replace_node(iff, cmovphi);
1343 return;
1344 }
1345
1346 // Now split the IF
1347 do_split_if(iff);
1348 return;
1349 }
1350
1351 // Two identical ifs back to back can be merged
1352 if (identical_backtoback_ifs(n) && can_split_if(n->in(0))) {
1353 Node *n_ctrl = n->in(0);
1354 PhiNode* bolphi = PhiNode::make_blank(n_ctrl, n->in(1));
1355 IfNode* dom_if = idom(n_ctrl)->as_If();
1356 Node* proj_true = dom_if->proj_out(1);
1357 Node* proj_false = dom_if->proj_out(0);
1358 Node* con_true = _igvn.makecon(TypeInt::ONE);
1359 Node* con_false = _igvn.makecon(TypeInt::ZERO);
1360
1361 for (uint i = 1; i < n_ctrl->req(); i++) {
1362 if (is_dominator(proj_true, n_ctrl->in(i))) {
1363 bolphi->init_req(i, con_true);
1364 } else {
1365 assert(is_dominator(proj_false, n_ctrl->in(i)), "bad if");
1366 bolphi->init_req(i, con_false);
1367 }
1368 }
1369 register_new_node(bolphi, n_ctrl);
1370 _igvn.replace_input_of(n, 1, bolphi);
1371
1372 // Now split the IF
1373 do_split_if(n);
1374 return;
1375 }
1376
1377 // Check for an IF ready to split; one that has its
1378 // condition codes input coming from a Phi at the block start.
1379 int n_op = n->Opcode();
1380
1381 // Check for an IF being dominated by another IF same test
1382 if (n_op == Op_If ||
1383 n_op == Op_RangeCheck) {
1384 Node *bol = n->in(1);
1385 uint max = bol->outcnt();
1386 // Check for same test used more than once?
1387 if (max > 1 && bol->is_Bool()) {
1388 // Search up IDOMs to see if this IF is dominated.
1389 Node *cutoff = get_ctrl(bol);
1390
1391 // Now search up IDOMs till cutoff, looking for a dominating test
1392 Node *prevdom = n;
1393 Node *dom = idom(prevdom);
1394 while (dom != cutoff) {
1395 if (dom->req() > 1 && dom->in(1) == bol && prevdom->in(0) == dom) {
1396 // It's invalid to move control dependent data nodes in the inner
1397 // strip-mined loop, because:
1398 // 1) break validation of LoopNode::verify_strip_mined()
1399 // 2) move code with side-effect in strip-mined loop
1400 // Move to the exit of outer strip-mined loop in that case.
1401 Node* out_le = is_inner_of_stripmined_loop(dom);
1402 if (out_le != NULL) {
1403 prevdom = out_le;
1404 }
1405 // Replace the dominated test with an obvious true or false.
1406 // Place it on the IGVN worklist for later cleanup.
1407 C->set_major_progress();
1408 dominated_by(prevdom, n, false, true);
1409 #ifndef PRODUCT
1410 if( VerifyLoopOptimizations ) verify();
1411 #endif
1412 return;
1413 }
1414 prevdom = dom;
1415 dom = idom(prevdom);
1416 }
1417 }
1418 }
1419
1420 try_sink_out_of_loop(n);
1421
1422 try_move_store_after_loop(n);
1423
1424 // Check for Opaque2's who's loop has disappeared - who's input is in the
1425 // same loop nest as their output. Remove 'em, they are no longer useful.
1426 if( n_op == Op_Opaque2 &&
1427 n->in(1) != NULL &&
1428 get_loop(get_ctrl(n)) == get_loop(get_ctrl(n->in(1))) ) {
1429 _igvn.replace_node( n, n->in(1) );
1430 }
1431 }
1432
1433 // See if a shared loop-varying computation has no loop-varying uses.
1434 // Happens if something is only used for JVM state in uncommon trap exits,
1435 // like various versions of induction variable+offset. Clone the
1436 // computation per usage to allow it to sink out of the loop.
1437 void PhaseIdealLoop::try_sink_out_of_loop(Node* n) {
1438 bool is_raw_to_oop_cast = n->is_ConstraintCast() &&
1439 n->in(1)->bottom_type()->isa_rawptr() &&
1440 !n->bottom_type()->isa_rawptr();
1441 if (has_ctrl(n) &&
1442 !n->is_Phi() &&
1443 !n->is_Bool() &&
1444 !n->is_Proj() &&
1445 !n->is_MergeMem() &&
1446 !n->is_CMove() &&
1447 !is_raw_to_oop_cast && // don't extend live ranges of raw oops
1448 n->Opcode() != Op_Opaque4 &&
1449 !n->is_Type()) {
1450 Node *n_ctrl = get_ctrl(n);
1451 IdealLoopTree *n_loop = get_loop(n_ctrl);
1452
1453 if (n->in(0) != NULL) {
1454 IdealLoopTree* loop_ctrl = get_loop(n->in(0));
1455 if (n_loop != loop_ctrl && n_loop->is_member(loop_ctrl)) {
1456 // n has a control input inside a loop but get_ctrl() is member of an outer loop. This could happen, for example,
1457 // for Div nodes inside a loop (control input inside loop) without a use except for an UCT (outside the loop).
1458 // Rewire control of n to right outside of the loop, regardless if its input(s) are later sunk or not.
1459 _igvn.replace_input_of(n, 0, place_outside_loop(n_ctrl, loop_ctrl));
1460 }
1461 }
1462 if (n_loop != _ltree_root && n->outcnt() > 1) {
1463 // Compute early control: needed for anti-dependence analysis. It's also possible that as a result of
1464 // previous transformations in this loop opts round, the node can be hoisted now: early control will tell us.
1465 Node* early_ctrl = compute_early_ctrl(n, n_ctrl);
1466 if (n_loop->is_member(get_loop(early_ctrl)) && // check that this one can't be hoisted now
1467 ctrl_of_all_uses_out_of_loop(n, early_ctrl, n_loop)) { // All uses in outer loops!
1468 assert(!n->is_Store() && !n->is_LoadStore(), "no node with a side effect");
1469 Node* outer_loop_clone = NULL;
1470 for (DUIterator_Last jmin, j = n->last_outs(jmin); j >= jmin;) {
1471 Node* u = n->last_out(j); // Clone private computation per use
1472 _igvn.rehash_node_delayed(u);
1473 Node* x = n->clone(); // Clone computation
1474 Node* x_ctrl = NULL;
1475 if (u->is_Phi()) {
1476 // Replace all uses of normal nodes. Replace Phi uses
1477 // individually, so the separate Nodes can sink down
1478 // different paths.
1479 uint k = 1;
1480 while (u->in(k) != n) k++;
1481 u->set_req(k, x);
1482 // x goes next to Phi input path
1483 x_ctrl = u->in(0)->in(k);
1484 // Find control for 'x' next to use but not inside inner loops.
1485 x_ctrl = place_outside_loop(x_ctrl, n_loop);
1486 --j;
1487 } else { // Normal use
1488 if (has_ctrl(u)) {
1489 x_ctrl = get_ctrl(u);
1490 } else {
1491 x_ctrl = u->in(0);
1492 }
1493 // Find control for 'x' next to use but not inside inner loops.
1494 x_ctrl = place_outside_loop(x_ctrl, n_loop);
1495 // Replace all uses
1496 if (u->is_ConstraintCast() && u->bottom_type()->higher_equal(_igvn.type(n)) && u->in(0) == x_ctrl) {
1497 // If we're sinking a chain of data nodes, we might have inserted a cast to pin the use which is not necessary
1498 // anymore now that we're going to pin n as well
1499 _igvn.replace_node(u, x);
1500 --j;
1501 } else {
1502 int nb = u->replace_edge(n, x, &_igvn);
1503 j -= nb;
1504 }
1505 }
1506
1507 if (n->is_Load()) {
1508 // For loads, add a control edge to a CFG node outside of the loop
1509 // to force them to not combine and return back inside the loop
1510 // during GVN optimization (4641526).
1511 assert(x_ctrl == get_late_ctrl_with_anti_dep(x->as_Load(), early_ctrl, x_ctrl), "anti-dependences were already checked");
1512
1513 IdealLoopTree* x_loop = get_loop(x_ctrl);
1514 Node* x_head = x_loop->_head;
1515 if (x_head->is_Loop() && x_head->is_OuterStripMinedLoop()) {
1516 // Do not add duplicate LoadNodes to the outer strip mined loop
1517 if (outer_loop_clone != NULL) {
1518 _igvn.replace_node(x, outer_loop_clone);
1519 continue;
1520 }
1521 outer_loop_clone = x;
1522 }
1523 x->set_req(0, x_ctrl);
1524 } else if (n->in(0) != NULL){
1525 x->set_req(0, x_ctrl);
1526 }
1527 assert(dom_depth(n_ctrl) <= dom_depth(x_ctrl), "n is later than its clone");
1528 assert(!n_loop->is_member(get_loop(x_ctrl)), "should have moved out of loop");
1529 register_new_node(x, x_ctrl);
1530
1531 // Chain of AddP: (AddP base (AddP base )) must keep the same base after sinking so:
1532 // 1- We don't add a CastPP here when the first one is sunk so if the second one is not, their bases remain
1533 // the same.
1534 // (see 2- below)
1535 assert(!x->is_AddP() || !x->in(AddPNode::Address)->is_AddP() ||
1536 x->in(AddPNode::Address)->in(AddPNode::Base) == x->in(AddPNode::Base) ||
1537 !x->in(AddPNode::Address)->in(AddPNode::Base)->eqv_uncast(x->in(AddPNode::Base)), "unexpected AddP shape");
1538 if (x->in(0) == NULL && !x->is_DecodeNarrowPtr() &&
1539 !(x->is_AddP() && x->in(AddPNode::Address)->is_AddP() && x->in(AddPNode::Address)->in(AddPNode::Base) == x->in(AddPNode::Base))) {
1540 assert(!x->is_Load(), "load should be pinned");
1541 // Use a cast node to pin clone out of loop
1542 Node* cast = NULL;
1543 for (uint k = 0; k < x->req(); k++) {
1544 Node* in = x->in(k);
1545 if (in != NULL && n_loop->is_member(get_loop(get_ctrl(in)))) {
1546 const Type* in_t = _igvn.type(in);
1547 cast = ConstraintCastNode::make_cast_for_type(x_ctrl, in, in_t, ConstraintCastNode::UnconditionalDependency);
1548 }
1549 if (cast != NULL) {
1550 register_new_node(cast, x_ctrl);
1551 x->replace_edge(in, cast);
1552 // Chain of AddP:
1553 // 2- A CastPP of the base is only added now that both AddP nodes are sunk
1554 if (x->is_AddP() && k == AddPNode::Base) {
1555 for (DUIterator_Fast imax, i = x->fast_outs(imax); i < imax; i++) {
1556 Node* u = x->fast_out(i);
1557 if (u->is_AddP() && u->in(AddPNode::Base) == n->in(AddPNode::Base)) {
1558 _igvn.replace_input_of(u, AddPNode::Base, cast);
1559 assert(u->find_out_with(Op_AddP) == NULL, "more than 2 chained AddP nodes?");
1560 }
1561 }
1562 }
1563 break;
1564 }
1565 }
1566 assert(cast != NULL, "must have added a cast to pin the node");
1567 }
1568 }
1569 _igvn.remove_dead_node(n);
1570 }
1571 _dom_lca_tags_round = 0;
1572 }
1573 }
1574 }
1575
1576 Node* PhaseIdealLoop::compute_early_ctrl(Node* n, Node* n_ctrl) {
1577 Node* early_ctrl = NULL;
1578 ResourceMark rm;
1579 Unique_Node_List wq;
1580 wq.push(n);
1581 for (uint i = 0; i < wq.size(); i++) {
1582 Node* m = wq.at(i);
1583 Node* c = NULL;
1584 if (m->is_CFG()) {
1585 c = m;
1586 } else if (m->pinned()) {
1587 c = m->in(0);
1588 } else {
1589 for (uint j = 0; j < m->req(); j++) {
1590 Node* in = m->in(j);
1591 if (in == NULL) {
1592 continue;
1593 }
1594 wq.push(in);
1595 }
1596 }
1597 if (c != NULL) {
1598 assert(is_dominator(c, n_ctrl), "");
1599 if (early_ctrl == NULL) {
1600 early_ctrl = c;
1601 } else if (is_dominator(early_ctrl, c)) {
1602 early_ctrl = c;
1603 }
1604 }
1605 }
1606 assert(is_dominator(early_ctrl, n_ctrl), "early control must dominate current control");
1607 return early_ctrl;
1608 }
1609
1610 bool PhaseIdealLoop::ctrl_of_all_uses_out_of_loop(const Node* n, Node* n_ctrl, IdealLoopTree* n_loop) {
1611 for (DUIterator_Fast imax, i = n->fast_outs(imax); i < imax; i++) {
1612 Node* u = n->fast_out(i);
1613 if (u->Opcode() == Op_Opaque1) {
1614 return false; // Found loop limit, bugfix for 4677003
1615 }
1616 // We can't reuse tags in PhaseIdealLoop::dom_lca_for_get_late_ctrl_internal() so make sure calls to
1617 // get_late_ctrl_with_anti_dep() use their own tag
1618 _dom_lca_tags_round++;
1619 assert(_dom_lca_tags_round != 0, "shouldn't wrap around");
1620
1621 if (u->is_Phi()) {
1622 for (uint j = 1; j < u->req(); ++j) {
1623 if (u->in(j) == n && !ctrl_of_use_out_of_loop(n, n_ctrl, n_loop, u->in(0)->in(j))) {
1624 return false;
1625 }
1626 }
1627 } else {
1628 Node* ctrl = has_ctrl(u) ? get_ctrl(u) : u->in(0);
1629 if (!ctrl_of_use_out_of_loop(n, n_ctrl, n_loop, ctrl)) {
1630 return false;
1631 }
1632 }
1633 }
1634 return true;
1635 }
1636
1637 bool PhaseIdealLoop::ctrl_of_use_out_of_loop(const Node* n, Node* n_ctrl, IdealLoopTree* n_loop, Node* ctrl) {
1638 if (n->is_Load()) {
1639 ctrl = get_late_ctrl_with_anti_dep(n->as_Load(), n_ctrl, ctrl);
1640 }
1641 IdealLoopTree *u_loop = get_loop(ctrl);
1642 if (u_loop == n_loop) {
1643 return false; // Found loop-varying use
1644 }
1645 if (n_loop->is_member(u_loop)) {
1646 return false; // Found use in inner loop
1647 }
1648 return true;
1649 }
1650
1651 //------------------------------split_if_with_blocks---------------------------
1652 // Check for aggressive application of 'split-if' optimization,
1653 // using basic block level info.
1654 void PhaseIdealLoop::split_if_with_blocks(VectorSet &visited, Node_Stack &nstack) {
1655 Node* root = C->root();
1656 visited.set(root->_idx); // first, mark root as visited
1657 // Do pre-visit work for root
1658 Node* n = split_if_with_blocks_pre(root);
1659 uint cnt = n->outcnt();
1660 uint i = 0;
1661
1662 while (true) {
1663 // Visit all children
1664 if (i < cnt) {
1665 Node* use = n->raw_out(i);
1666 ++i;
1667 if (use->outcnt() != 0 && !visited.test_set(use->_idx)) {
1668 // Now do pre-visit work for this use
1669 use = split_if_with_blocks_pre(use);
1670 nstack.push(n, i); // Save parent and next use's index.
1671 n = use; // Process all children of current use.
1672 cnt = use->outcnt();
1673 i = 0;
1674 }
1675 }
1676 else {
1677 // All of n's children have been processed, complete post-processing.
1678 if (cnt != 0 && !n->is_Con()) {
1679 assert(has_node(n), "no dead nodes");
1680 split_if_with_blocks_post(n);
1681 }
1682 if (must_throttle_split_if()) {
1683 nstack.clear();
1684 }
1685 if (nstack.is_empty()) {
1686 // Finished all nodes on stack.
1687 break;
1688 }
1689 // Get saved parent node and next use's index. Visit the rest of uses.
1690 n = nstack.node();
1691 cnt = n->outcnt();
1692 i = nstack.index();
1693 nstack.pop();
1694 }
1695 }
1696 }
1697
1698
1699 //=============================================================================
1700 //
1701 // C L O N E A L O O P B O D Y
1702 //
1703
1704 //------------------------------clone_iff--------------------------------------
1705 // Passed in a Phi merging (recursively) some nearly equivalent Bool/Cmps.
1706 // "Nearly" because all Nodes have been cloned from the original in the loop,
1707 // but the fall-in edges to the Cmp are different. Clone bool/Cmp pairs
1708 // through the Phi recursively, and return a Bool.
1709 Node* PhaseIdealLoop::clone_iff(PhiNode *phi, IdealLoopTree *loop) {
1710
1711 // Convert this Phi into a Phi merging Bools
1712 uint i;
1713 for (i = 1; i < phi->req(); i++) {
1714 Node *b = phi->in(i);
1715 if (b->is_Phi()) {
1716 _igvn.replace_input_of(phi, i, clone_iff(b->as_Phi(), loop));
1717 } else {
1718 assert(b->is_Bool() || b->Opcode() == Op_Opaque4, "");
1719 }
1720 }
1721
1722 Node* n = phi->in(1);
1723 Node* sample_opaque = NULL;
1724 Node *sample_bool = NULL;
1725 if (n->Opcode() == Op_Opaque4) {
1726 sample_opaque = n;
1727 sample_bool = n->in(1);
1728 assert(sample_bool->is_Bool(), "wrong type");
1729 } else {
1730 sample_bool = n;
1731 }
1732 Node *sample_cmp = sample_bool->in(1);
1733
1734 // Make Phis to merge the Cmp's inputs.
1735 PhiNode *phi1 = new PhiNode(phi->in(0), Type::TOP);
1736 PhiNode *phi2 = new PhiNode(phi->in(0), Type::TOP);
1737 for (i = 1; i < phi->req(); i++) {
1738 Node *n1 = sample_opaque == NULL ? phi->in(i)->in(1)->in(1) : phi->in(i)->in(1)->in(1)->in(1);
1739 Node *n2 = sample_opaque == NULL ? phi->in(i)->in(1)->in(2) : phi->in(i)->in(1)->in(1)->in(2);
1740 phi1->set_req(i, n1);
1741 phi2->set_req(i, n2);
1742 phi1->set_type(phi1->type()->meet_speculative(n1->bottom_type()));
1743 phi2->set_type(phi2->type()->meet_speculative(n2->bottom_type()));
1744 }
1745 // See if these Phis have been made before.
1746 // Register with optimizer
1747 Node *hit1 = _igvn.hash_find_insert(phi1);
1748 if (hit1) { // Hit, toss just made Phi
1749 _igvn.remove_dead_node(phi1); // Remove new phi
1750 assert(hit1->is_Phi(), "" );
1751 phi1 = (PhiNode*)hit1; // Use existing phi
1752 } else { // Miss
1753 _igvn.register_new_node_with_optimizer(phi1);
1754 }
1755 Node *hit2 = _igvn.hash_find_insert(phi2);
1756 if (hit2) { // Hit, toss just made Phi
1757 _igvn.remove_dead_node(phi2); // Remove new phi
1758 assert(hit2->is_Phi(), "" );
1759 phi2 = (PhiNode*)hit2; // Use existing phi
1760 } else { // Miss
1761 _igvn.register_new_node_with_optimizer(phi2);
1762 }
1763 // Register Phis with loop/block info
1764 set_ctrl(phi1, phi->in(0));
1765 set_ctrl(phi2, phi->in(0));
1766 // Make a new Cmp
1767 Node *cmp = sample_cmp->clone();
1768 cmp->set_req(1, phi1);
1769 cmp->set_req(2, phi2);
1770 _igvn.register_new_node_with_optimizer(cmp);
1771 set_ctrl(cmp, phi->in(0));
1772
1773 // Make a new Bool
1774 Node *b = sample_bool->clone();
1775 b->set_req(1,cmp);
1776 _igvn.register_new_node_with_optimizer(b);
1777 set_ctrl(b, phi->in(0));
1778
1779 if (sample_opaque != NULL) {
1780 Node* opaque = sample_opaque->clone();
1781 opaque->set_req(1, b);
1782 _igvn.register_new_node_with_optimizer(opaque);
1783 set_ctrl(opaque, phi->in(0));
1784 return opaque;
1785 }
1786
1787 assert(b->is_Bool(), "");
1788 return b;
1789 }
1790
1791 //------------------------------clone_bool-------------------------------------
1792 // Passed in a Phi merging (recursively) some nearly equivalent Bool/Cmps.
1793 // "Nearly" because all Nodes have been cloned from the original in the loop,
1794 // but the fall-in edges to the Cmp are different. Clone bool/Cmp pairs
1795 // through the Phi recursively, and return a Bool.
1796 CmpNode *PhaseIdealLoop::clone_bool( PhiNode *phi, IdealLoopTree *loop ) {
1797 uint i;
1798 // Convert this Phi into a Phi merging Bools
1799 for( i = 1; i < phi->req(); i++ ) {
1800 Node *b = phi->in(i);
1801 if( b->is_Phi() ) {
1802 _igvn.replace_input_of(phi, i, clone_bool( b->as_Phi(), loop ));
1803 } else {
1804 assert( b->is_Cmp() || b->is_top(), "inputs are all Cmp or TOP" );
1805 }
1806 }
1807
1808 Node *sample_cmp = phi->in(1);
1809
1810 // Make Phis to merge the Cmp's inputs.
1811 PhiNode *phi1 = new PhiNode( phi->in(0), Type::TOP );
1812 PhiNode *phi2 = new PhiNode( phi->in(0), Type::TOP );
1813 for( uint j = 1; j < phi->req(); j++ ) {
1814 Node *cmp_top = phi->in(j); // Inputs are all Cmp or TOP
1815 Node *n1, *n2;
1816 if( cmp_top->is_Cmp() ) {
1817 n1 = cmp_top->in(1);
1818 n2 = cmp_top->in(2);
1819 } else {
1820 n1 = n2 = cmp_top;
1821 }
1822 phi1->set_req( j, n1 );
1823 phi2->set_req( j, n2 );
1824 phi1->set_type(phi1->type()->meet_speculative(n1->bottom_type()));
1825 phi2->set_type(phi2->type()->meet_speculative(n2->bottom_type()));
1826 }
1827
1828 // See if these Phis have been made before.
1829 // Register with optimizer
1830 Node *hit1 = _igvn.hash_find_insert(phi1);
1831 if( hit1 ) { // Hit, toss just made Phi
1832 _igvn.remove_dead_node(phi1); // Remove new phi
1833 assert( hit1->is_Phi(), "" );
1834 phi1 = (PhiNode*)hit1; // Use existing phi
1835 } else { // Miss
1836 _igvn.register_new_node_with_optimizer(phi1);
1837 }
1838 Node *hit2 = _igvn.hash_find_insert(phi2);
1839 if( hit2 ) { // Hit, toss just made Phi
1840 _igvn.remove_dead_node(phi2); // Remove new phi
1841 assert( hit2->is_Phi(), "" );
1842 phi2 = (PhiNode*)hit2; // Use existing phi
1843 } else { // Miss
1844 _igvn.register_new_node_with_optimizer(phi2);
1845 }
1846 // Register Phis with loop/block info
1847 set_ctrl(phi1, phi->in(0));
1848 set_ctrl(phi2, phi->in(0));
1849 // Make a new Cmp
1850 Node *cmp = sample_cmp->clone();
1851 cmp->set_req( 1, phi1 );
1852 cmp->set_req( 2, phi2 );
1853 _igvn.register_new_node_with_optimizer(cmp);
1854 set_ctrl(cmp, phi->in(0));
1855
1856 assert( cmp->is_Cmp(), "" );
1857 return (CmpNode*)cmp;
1858 }
1859
1860 //------------------------------sink_use---------------------------------------
1861 // If 'use' was in the loop-exit block, it now needs to be sunk
1862 // below the post-loop merge point.
1863 void PhaseIdealLoop::sink_use( Node *use, Node *post_loop ) {
1864 if (!use->is_CFG() && get_ctrl(use) == post_loop->in(2)) {
1865 set_ctrl(use, post_loop);
1866 for (DUIterator j = use->outs(); use->has_out(j); j++)
1867 sink_use(use->out(j), post_loop);
1868 }
1869 }
1870
1871 void PhaseIdealLoop::clone_loop_handle_data_uses(Node* old, Node_List &old_new,
1872 IdealLoopTree* loop, IdealLoopTree* outer_loop,
1873 Node_List*& split_if_set, Node_List*& split_bool_set,
1874 Node_List*& split_cex_set, Node_List& worklist,
1875 uint new_counter, CloneLoopMode mode) {
1876 Node* nnn = old_new[old->_idx];
1877 // Copy uses to a worklist, so I can munge the def-use info
1878 // with impunity.
1879 for (DUIterator_Fast jmax, j = old->fast_outs(jmax); j < jmax; j++)
1880 worklist.push(old->fast_out(j));
1881
1882 while( worklist.size() ) {
1883 Node *use = worklist.pop();
1884 if (!has_node(use)) continue; // Ignore dead nodes
1885 if (use->in(0) == C->top()) continue;
1886 IdealLoopTree *use_loop = get_loop( has_ctrl(use) ? get_ctrl(use) : use );
1887 // Check for data-use outside of loop - at least one of OLD or USE
1888 // must not be a CFG node.
1889 #ifdef ASSERT
1890 if (loop->_head->as_Loop()->is_strip_mined() && outer_loop->is_member(use_loop) && !loop->is_member(use_loop) && old_new[use->_idx] == NULL) {
1891 Node* sfpt = loop->_head->as_CountedLoop()->outer_safepoint();
1892 assert(mode != IgnoreStripMined, "incorrect cloning mode");
1893 assert((mode == ControlAroundStripMined && use == sfpt) || !use->is_reachable_from_root(), "missed a node");
1894 }
1895 #endif
1896 if (!loop->is_member(use_loop) && !outer_loop->is_member(use_loop) && (!old->is_CFG() || !use->is_CFG())) {
1897
1898 // If the Data use is an IF, that means we have an IF outside of the
1899 // loop that is switching on a condition that is set inside of the
1900 // loop. Happens if people set a loop-exit flag; then test the flag
1901 // in the loop to break the loop, then test is again outside of the
1902 // loop to determine which way the loop exited.
1903 // Loop predicate If node connects to Bool node through Opaque1 node.
1904 if (use->is_If() || use->is_CMove() || C->is_predicate_opaq(use) || use->Opcode() == Op_Opaque4) {
1905 // Since this code is highly unlikely, we lazily build the worklist
1906 // of such Nodes to go split.
1907 if (!split_if_set) {
1908 split_if_set = new Node_List();
1909 }
1910 split_if_set->push(use);
1911 }
1912 if (use->is_Bool()) {
1913 if (!split_bool_set) {
1914 split_bool_set = new Node_List();
1915 }
1916 split_bool_set->push(use);
1917 }
1918 if (use->Opcode() == Op_CreateEx) {
1919 if (!split_cex_set) {
1920 split_cex_set = new Node_List();
1921 }
1922 split_cex_set->push(use);
1923 }
1924
1925
1926 // Get "block" use is in
1927 uint idx = 0;
1928 while( use->in(idx) != old ) idx++;
1929 Node *prev = use->is_CFG() ? use : get_ctrl(use);
1930 assert(!loop->is_member(get_loop(prev)) && !outer_loop->is_member(get_loop(prev)), "" );
1931 Node *cfg = prev->_idx >= new_counter
1932 ? prev->in(2)
1933 : idom(prev);
1934 if( use->is_Phi() ) // Phi use is in prior block
1935 cfg = prev->in(idx); // NOT in block of Phi itself
1936 if (cfg->is_top()) { // Use is dead?
1937 _igvn.replace_input_of(use, idx, C->top());
1938 continue;
1939 }
1940
1941 // If use is referenced through control edge... (idx == 0)
1942 if (mode == IgnoreStripMined && idx == 0) {
1943 LoopNode *head = loop->_head->as_Loop();
1944 if (head->is_strip_mined() && is_dominator(head->outer_loop_exit(), prev)) {
1945 // That node is outside the inner loop, leave it outside the
1946 // outer loop as well to not confuse verification code.
1947 assert(!loop->_parent->is_member(use_loop), "should be out of the outer loop");
1948 _igvn.replace_input_of(use, 0, head->outer_loop_exit());
1949 continue;
1950 }
1951 }
1952
1953 while(!outer_loop->is_member(get_loop(cfg))) {
1954 prev = cfg;
1955 cfg = cfg->_idx >= new_counter ? cfg->in(2) : idom(cfg);
1956 }
1957 // If the use occurs after merging several exits from the loop, then
1958 // old value must have dominated all those exits. Since the same old
1959 // value was used on all those exits we did not need a Phi at this
1960 // merge point. NOW we do need a Phi here. Each loop exit value
1961 // is now merged with the peeled body exit; each exit gets its own
1962 // private Phi and those Phis need to be merged here.
1963 Node *phi;
1964 if( prev->is_Region() ) {
1965 if( idx == 0 ) { // Updating control edge?
1966 phi = prev; // Just use existing control
1967 } else { // Else need a new Phi
1968 phi = PhiNode::make( prev, old );
1969 // Now recursively fix up the new uses of old!
1970 for( uint i = 1; i < prev->req(); i++ ) {
1971 worklist.push(phi); // Onto worklist once for each 'old' input
1972 }
1973 }
1974 } else {
1975 // Get new RegionNode merging old and new loop exits
1976 prev = old_new[prev->_idx];
1977 assert( prev, "just made this in step 7" );
1978 if( idx == 0) { // Updating control edge?
1979 phi = prev; // Just use existing control
1980 } else { // Else need a new Phi
1981 // Make a new Phi merging data values properly
1982 phi = PhiNode::make( prev, old );
1983 phi->set_req( 1, nnn );
1984 }
1985 }
1986 // If inserting a new Phi, check for prior hits
1987 if( idx != 0 ) {
1988 Node *hit = _igvn.hash_find_insert(phi);
1989 if( hit == NULL ) {
1990 _igvn.register_new_node_with_optimizer(phi); // Register new phi
1991 } else { // or
1992 // Remove the new phi from the graph and use the hit
1993 _igvn.remove_dead_node(phi);
1994 phi = hit; // Use existing phi
1995 }
1996 set_ctrl(phi, prev);
1997 }
1998 // Make 'use' use the Phi instead of the old loop body exit value
1999 _igvn.replace_input_of(use, idx, phi);
2000 if( use->_idx >= new_counter ) { // If updating new phis
2001 // Not needed for correctness, but prevents a weak assert
2002 // in AddPNode from tripping (when we end up with different
2003 // base & derived Phis that will become the same after
2004 // IGVN does CSE).
2005 Node *hit = _igvn.hash_find_insert(use);
2006 if( hit ) // Go ahead and re-hash for hits.
2007 _igvn.replace_node( use, hit );
2008 }
2009
2010 // If 'use' was in the loop-exit block, it now needs to be sunk
2011 // below the post-loop merge point.
2012 sink_use( use, prev );
2013 }
2014 }
2015 }
2016
2017 static void clone_outer_loop_helper(Node* n, const IdealLoopTree *loop, const IdealLoopTree* outer_loop,
2018 const Node_List &old_new, Unique_Node_List& wq, PhaseIdealLoop* phase,
2019 bool check_old_new) {
2020 for (DUIterator_Fast jmax, j = n->fast_outs(jmax); j < jmax; j++) {
2021 Node* u = n->fast_out(j);
2022 assert(check_old_new || old_new[u->_idx] == NULL, "shouldn't have been cloned");
2023 if (!u->is_CFG() && (!check_old_new || old_new[u->_idx] == NULL)) {
2024 Node* c = phase->get_ctrl(u);
2025 IdealLoopTree* u_loop = phase->get_loop(c);
2026 assert(!loop->is_member(u_loop), "can be in outer loop or out of both loops only");
2027 if (outer_loop->is_member(u_loop) ||
2028 // nodes pinned with control in the outer loop but not referenced from the safepoint must be moved out of
2029 // the outer loop too
2030 (u->in(0) != NULL && outer_loop->is_member(phase->get_loop(u->in(0))))) {
2031 wq.push(u);
2032 }
2033 }
2034 }
2035 }
2036
2037 void PhaseIdealLoop::clone_outer_loop(LoopNode* head, CloneLoopMode mode, IdealLoopTree *loop,
2038 IdealLoopTree* outer_loop, int dd, Node_List &old_new,
2039 Node_List& extra_data_nodes) {
2040 if (head->is_strip_mined() && mode != IgnoreStripMined) {
2041 CountedLoopNode* cl = head->as_CountedLoop();
2042 Node* l = cl->outer_loop();
2043 Node* tail = cl->outer_loop_tail();
2044 IfNode* le = cl->outer_loop_end();
2045 Node* sfpt = cl->outer_safepoint();
2046 CountedLoopEndNode* cle = cl->loopexit();
2047 CountedLoopNode* new_cl = old_new[cl->_idx]->as_CountedLoop();
2048 CountedLoopEndNode* new_cle = new_cl->as_CountedLoop()->loopexit_or_null();
2049 Node* cle_out = cle->proj_out(false);
2050
2051 Node* new_sfpt = NULL;
2052 Node* new_cle_out = cle_out->clone();
2053 old_new.map(cle_out->_idx, new_cle_out);
2054 if (mode == CloneIncludesStripMined) {
2055 // clone outer loop body
2056 Node* new_l = l->clone();
2057 Node* new_tail = tail->clone();
2058 IfNode* new_le = le->clone()->as_If();
2059 new_sfpt = sfpt->clone();
2060
2061 set_loop(new_l, outer_loop->_parent);
2062 set_idom(new_l, new_l->in(LoopNode::EntryControl), dd);
2063 set_loop(new_cle_out, outer_loop->_parent);
2064 set_idom(new_cle_out, new_cle, dd);
2065 set_loop(new_sfpt, outer_loop->_parent);
2066 set_idom(new_sfpt, new_cle_out, dd);
2067 set_loop(new_le, outer_loop->_parent);
2068 set_idom(new_le, new_sfpt, dd);
2069 set_loop(new_tail, outer_loop->_parent);
2070 set_idom(new_tail, new_le, dd);
2071 set_idom(new_cl, new_l, dd);
2072
2073 old_new.map(l->_idx, new_l);
2074 old_new.map(tail->_idx, new_tail);
2075 old_new.map(le->_idx, new_le);
2076 old_new.map(sfpt->_idx, new_sfpt);
2077
2078 new_l->set_req(LoopNode::LoopBackControl, new_tail);
2079 new_l->set_req(0, new_l);
2080 new_tail->set_req(0, new_le);
2081 new_le->set_req(0, new_sfpt);
2082 new_sfpt->set_req(0, new_cle_out);
2083 new_cle_out->set_req(0, new_cle);
2084 new_cl->set_req(LoopNode::EntryControl, new_l);
2085
2086 _igvn.register_new_node_with_optimizer(new_l);
2087 _igvn.register_new_node_with_optimizer(new_tail);
2088 _igvn.register_new_node_with_optimizer(new_le);
2089 } else {
2090 Node *newhead = old_new[loop->_head->_idx];
2091 newhead->as_Loop()->clear_strip_mined();
2092 _igvn.replace_input_of(newhead, LoopNode::EntryControl, newhead->in(LoopNode::EntryControl)->in(LoopNode::EntryControl));
2093 set_idom(newhead, newhead->in(LoopNode::EntryControl), dd);
2094 }
2095 // Look at data node that were assigned a control in the outer
2096 // loop: they are kept in the outer loop by the safepoint so start
2097 // from the safepoint node's inputs.
2098 IdealLoopTree* outer_loop = get_loop(l);
2099 Node_Stack stack(2);
2100 stack.push(sfpt, 1);
2101 uint new_counter = C->unique();
2102 while (stack.size() > 0) {
2103 Node* n = stack.node();
2104 uint i = stack.index();
2105 while (i < n->req() &&
2106 (n->in(i) == NULL ||
2107 !has_ctrl(n->in(i)) ||
2108 get_loop(get_ctrl(n->in(i))) != outer_loop ||
2109 (old_new[n->in(i)->_idx] != NULL && old_new[n->in(i)->_idx]->_idx >= new_counter))) {
2110 i++;
2111 }
2112 if (i < n->req()) {
2113 stack.set_index(i+1);
2114 stack.push(n->in(i), 0);
2115 } else {
2116 assert(old_new[n->_idx] == NULL || n == sfpt || old_new[n->_idx]->_idx < new_counter, "no clone yet");
2117 Node* m = n == sfpt ? new_sfpt : n->clone();
2118 if (m != NULL) {
2119 for (uint i = 0; i < n->req(); i++) {
2120 if (m->in(i) != NULL && old_new[m->in(i)->_idx] != NULL) {
2121 m->set_req(i, old_new[m->in(i)->_idx]);
2122 }
2123 }
2124 } else {
2125 assert(n == sfpt && mode != CloneIncludesStripMined, "where's the safepoint clone?");
2126 }
2127 if (n != sfpt) {
2128 extra_data_nodes.push(n);
2129 _igvn.register_new_node_with_optimizer(m);
2130 assert(get_ctrl(n) == cle_out, "what other control?");
2131 set_ctrl(m, new_cle_out);
2132 old_new.map(n->_idx, m);
2133 }
2134 stack.pop();
2135 }
2136 }
2137 if (mode == CloneIncludesStripMined) {
2138 _igvn.register_new_node_with_optimizer(new_sfpt);
2139 _igvn.register_new_node_with_optimizer(new_cle_out);
2140 }
2141 // Some other transformation may have pessimistically assign some
2142 // data nodes to the outer loop. Set their control so they are out
2143 // of the outer loop.
2144 ResourceMark rm;
2145 Unique_Node_List wq;
2146 for (uint i = 0; i < extra_data_nodes.size(); i++) {
2147 Node* old = extra_data_nodes.at(i);
2148 clone_outer_loop_helper(old, loop, outer_loop, old_new, wq, this, true);
2149 }
2150
2151 Node* inner_out = sfpt->in(0);
2152 if (inner_out->outcnt() > 1) {
2153 clone_outer_loop_helper(inner_out, loop, outer_loop, old_new, wq, this, true);
2154 }
2155
2156 Node* new_ctrl = cl->outer_loop_exit();
2157 assert(get_loop(new_ctrl) != outer_loop, "must be out of the loop nest");
2158 for (uint i = 0; i < wq.size(); i++) {
2159 Node* n = wq.at(i);
2160 set_ctrl(n, new_ctrl);
2161 if (n->in(0) != NULL) {
2162 _igvn.replace_input_of(n, 0, new_ctrl);
2163 }
2164 clone_outer_loop_helper(n, loop, outer_loop, old_new, wq, this, false);
2165 }
2166 } else {
2167 Node *newhead = old_new[loop->_head->_idx];
2168 set_idom(newhead, newhead->in(LoopNode::EntryControl), dd);
2169 }
2170 }
2171
2172 //------------------------------clone_loop-------------------------------------
2173 //
2174 // C L O N E A L O O P B O D Y
2175 //
2176 // This is the basic building block of the loop optimizations. It clones an
2177 // entire loop body. It makes an old_new loop body mapping; with this mapping
2178 // you can find the new-loop equivalent to an old-loop node. All new-loop
2179 // nodes are exactly equal to their old-loop counterparts, all edges are the
2180 // same. All exits from the old-loop now have a RegionNode that merges the
2181 // equivalent new-loop path. This is true even for the normal "loop-exit"
2182 // condition. All uses of loop-invariant old-loop values now come from (one
2183 // or more) Phis that merge their new-loop equivalents.
2184 //
2185 // This operation leaves the graph in an illegal state: there are two valid
2186 // control edges coming from the loop pre-header to both loop bodies. I'll
2187 // definitely have to hack the graph after running this transform.
2188 //
2189 // From this building block I will further edit edges to perform loop peeling
2190 // or loop unrolling or iteration splitting (Range-Check-Elimination), etc.
2191 //
2192 // Parameter side_by_size_idom:
2193 // When side_by_size_idom is NULL, the dominator tree is constructed for
2194 // the clone loop to dominate the original. Used in construction of
2195 // pre-main-post loop sequence.
2196 // When nonnull, the clone and original are side-by-side, both are
2197 // dominated by the side_by_side_idom node. Used in construction of
2198 // unswitched loops.
2199 void PhaseIdealLoop::clone_loop( IdealLoopTree *loop, Node_List &old_new, int dd,
2200 CloneLoopMode mode, Node* side_by_side_idom) {
2201
2202 LoopNode* head = loop->_head->as_Loop();
2203 head->verify_strip_mined(1);
2204
2205 if (C->do_vector_loop() && PrintOpto) {
2206 const char* mname = C->method()->name()->as_quoted_ascii();
2207 if (mname != NULL) {
2208 tty->print("PhaseIdealLoop::clone_loop: for vectorize method %s\n", mname);
2209 }
2210 }
2211
2212 CloneMap& cm = C->clone_map();
2213 Dict* dict = cm.dict();
2214 if (C->do_vector_loop()) {
2215 cm.set_clone_idx(cm.max_gen()+1);
2216 #ifndef PRODUCT
2217 if (PrintOpto) {
2218 tty->print_cr("PhaseIdealLoop::clone_loop: _clone_idx %d", cm.clone_idx());
2219 loop->dump_head();
2220 }
2221 #endif
2222 }
2223
2224 // Step 1: Clone the loop body. Make the old->new mapping.
2225 uint i;
2226 for (i = 0; i < loop->_body.size(); i++) {
2227 Node* old = loop->_body.at(i);
2228 Node* nnn = old->clone();
2229 old_new.map(old->_idx, nnn);
2230 if (old->is_reduction()) {
2231 // Reduction flag is not copied by default. Copy it here when cloning the entire loop body.
2232 nnn->add_flag(Node::Flag_is_reduction);
2233 }
2234 if (C->do_vector_loop()) {
2235 cm.verify_insert_and_clone(old, nnn, cm.clone_idx());
2236 }
2237 _igvn.register_new_node_with_optimizer(nnn);
2238 }
2239
2240 IdealLoopTree* outer_loop = (head->is_strip_mined() && mode != IgnoreStripMined) ? get_loop(head->as_CountedLoop()->outer_loop()) : loop;
2241
2242 // Step 2: Fix the edges in the new body. If the old input is outside the
2243 // loop use it. If the old input is INside the loop, use the corresponding
2244 // new node instead.
2245 for( i = 0; i < loop->_body.size(); i++ ) {
2246 Node *old = loop->_body.at(i);
2247 Node *nnn = old_new[old->_idx];
2248 // Fix CFG/Loop controlling the new node
2249 if (has_ctrl(old)) {
2250 set_ctrl(nnn, old_new[get_ctrl(old)->_idx]);
2251 } else {
2252 set_loop(nnn, outer_loop->_parent);
2253 if (old->outcnt() > 0) {
2254 set_idom( nnn, old_new[idom(old)->_idx], dd );
2255 }
2256 }
2257 // Correct edges to the new node
2258 for( uint j = 0; j < nnn->req(); j++ ) {
2259 Node *n = nnn->in(j);
2260 if( n ) {
2261 IdealLoopTree *old_in_loop = get_loop( has_ctrl(n) ? get_ctrl(n) : n );
2262 if( loop->is_member( old_in_loop ) )
2263 nnn->set_req(j, old_new[n->_idx]);
2264 }
2265 }
2266 _igvn.hash_find_insert(nnn);
2267 }
2268
2269 Node_List extra_data_nodes; // data nodes in the outer strip mined loop
2270 clone_outer_loop(head, mode, loop, outer_loop, dd, old_new, extra_data_nodes);
2271
2272 // Step 3: Now fix control uses. Loop varying control uses have already
2273 // been fixed up (as part of all input edges in Step 2). Loop invariant
2274 // control uses must be either an IfFalse or an IfTrue. Make a merge
2275 // point to merge the old and new IfFalse/IfTrue nodes; make the use
2276 // refer to this.
2277 Node_List worklist;
2278 uint new_counter = C->unique();
2279 for( i = 0; i < loop->_body.size(); i++ ) {
2280 Node* old = loop->_body.at(i);
2281 if( !old->is_CFG() ) continue;
2282
2283 // Copy uses to a worklist, so I can munge the def-use info
2284 // with impunity.
2285 for (DUIterator_Fast jmax, j = old->fast_outs(jmax); j < jmax; j++)
2286 worklist.push(old->fast_out(j));
2287
2288 while( worklist.size() ) { // Visit all uses
2289 Node *use = worklist.pop();
2290 if (!has_node(use)) continue; // Ignore dead nodes
2291 IdealLoopTree *use_loop = get_loop( has_ctrl(use) ? get_ctrl(use) : use );
2292 if( !loop->is_member( use_loop ) && use->is_CFG() ) {
2293 // Both OLD and USE are CFG nodes here.
2294 assert( use->is_Proj(), "" );
2295 Node* nnn = old_new[old->_idx];
2296
2297 Node* newuse = NULL;
2298 if (head->is_strip_mined() && mode != IgnoreStripMined) {
2299 CountedLoopNode* cl = head->as_CountedLoop();
2300 CountedLoopEndNode* cle = cl->loopexit();
2301 Node* cle_out = cle->proj_out_or_null(false);
2302 if (use == cle_out) {
2303 IfNode* le = cl->outer_loop_end();
2304 use = le->proj_out(false);
2305 use_loop = get_loop(use);
2306 if (mode == CloneIncludesStripMined) {
2307 nnn = old_new[le->_idx];
2308 } else {
2309 newuse = old_new[cle_out->_idx];
2310 }
2311 }
2312 }
2313 if (newuse == NULL) {
2314 newuse = use->clone();
2315 }
2316
2317 // Clone the loop exit control projection
2318 if (C->do_vector_loop()) {
2319 cm.verify_insert_and_clone(use, newuse, cm.clone_idx());
2320 }
2321 newuse->set_req(0,nnn);
2322 _igvn.register_new_node_with_optimizer(newuse);
2323 set_loop(newuse, use_loop);
2324 set_idom(newuse, nnn, dom_depth(nnn) + 1 );
2325
2326 // We need a Region to merge the exit from the peeled body and the
2327 // exit from the old loop body.
2328 RegionNode *r = new RegionNode(3);
2329 // Map the old use to the new merge point
2330 old_new.map( use->_idx, r );
2331 uint dd_r = MIN2(dom_depth(newuse),dom_depth(use));
2332 assert( dd_r >= dom_depth(dom_lca(newuse,use)), "" );
2333
2334 // The original user of 'use' uses 'r' instead.
2335 for (DUIterator_Last lmin, l = use->last_outs(lmin); l >= lmin;) {
2336 Node* useuse = use->last_out(l);
2337 _igvn.rehash_node_delayed(useuse);
2338 uint uses_found = 0;
2339 if( useuse->in(0) == use ) {
2340 useuse->set_req(0, r);
2341 uses_found++;
2342 if( useuse->is_CFG() ) {
2343 // This is not a dom_depth > dd_r because when new
2344 // control flow is constructed by a loop opt, a node and
2345 // its dominator can end up at the same dom_depth
2346 assert(dom_depth(useuse) >= dd_r, "");
2347 set_idom(useuse, r, dom_depth(useuse));
2348 }
2349 }
2350 for( uint k = 1; k < useuse->req(); k++ ) {
2351 if( useuse->in(k) == use ) {
2352 useuse->set_req(k, r);
2353 uses_found++;
2354 if (useuse->is_Loop() && k == LoopNode::EntryControl) {
2355 // This is not a dom_depth > dd_r because when new
2356 // control flow is constructed by a loop opt, a node
2357 // and its dominator can end up at the same dom_depth
2358 assert(dom_depth(useuse) >= dd_r , "");
2359 set_idom(useuse, r, dom_depth(useuse));
2360 }
2361 }
2362 }
2363 l -= uses_found; // we deleted 1 or more copies of this edge
2364 }
2365
2366 // Now finish up 'r'
2367 r->set_req( 1, newuse );
2368 r->set_req( 2, use );
2369 _igvn.register_new_node_with_optimizer(r);
2370 set_loop(r, use_loop);
2371 set_idom(r, !side_by_side_idom ? newuse->in(0) : side_by_side_idom, dd_r);
2372 } // End of if a loop-exit test
2373 }
2374 }
2375
2376 // Step 4: If loop-invariant use is not control, it must be dominated by a
2377 // loop exit IfFalse/IfTrue. Find "proper" loop exit. Make a Region
2378 // there if needed. Make a Phi there merging old and new used values.
2379 Node_List *split_if_set = NULL;
2380 Node_List *split_bool_set = NULL;
2381 Node_List *split_cex_set = NULL;
2382 for( i = 0; i < loop->_body.size(); i++ ) {
2383 Node* old = loop->_body.at(i);
2384 clone_loop_handle_data_uses(old, old_new, loop, outer_loop, split_if_set,
2385 split_bool_set, split_cex_set, worklist, new_counter,
2386 mode);
2387 }
2388
2389 for (i = 0; i < extra_data_nodes.size(); i++) {
2390 Node* old = extra_data_nodes.at(i);
2391 clone_loop_handle_data_uses(old, old_new, loop, outer_loop, split_if_set,
2392 split_bool_set, split_cex_set, worklist, new_counter,
2393 mode);
2394 }
2395
2396 // Check for IFs that need splitting/cloning. Happens if an IF outside of
2397 // the loop uses a condition set in the loop. The original IF probably
2398 // takes control from one or more OLD Regions (which in turn get from NEW
2399 // Regions). In any case, there will be a set of Phis for each merge point
2400 // from the IF up to where the original BOOL def exists the loop.
2401 if (split_if_set) {
2402 while (split_if_set->size()) {
2403 Node *iff = split_if_set->pop();
2404 if (iff->in(1)->is_Phi()) {
2405 Node *b = clone_iff(iff->in(1)->as_Phi(), loop);
2406 _igvn.replace_input_of(iff, 1, b);
2407 }
2408 }
2409 }
2410 if (split_bool_set) {
2411 while (split_bool_set->size()) {
2412 Node *b = split_bool_set->pop();
2413 Node *phi = b->in(1);
2414 assert(phi->is_Phi(), "");
2415 CmpNode *cmp = clone_bool((PhiNode*)phi, loop);
2416 _igvn.replace_input_of(b, 1, cmp);
2417 }
2418 }
2419 if (split_cex_set) {
2420 while (split_cex_set->size()) {
2421 Node *b = split_cex_set->pop();
2422 assert(b->in(0)->is_Region(), "");
2423 assert(b->in(1)->is_Phi(), "");
2424 assert(b->in(0)->in(0) == b->in(1)->in(0), "");
2425 split_up(b, b->in(0), NULL);
2426 }
2427 }
2428
2429 }
2430
2431
2432 //---------------------- stride_of_possible_iv -------------------------------------
2433 // Looks for an iff/bool/comp with one operand of the compare
2434 // being a cycle involving an add and a phi,
2435 // with an optional truncation (left-shift followed by a right-shift)
2436 // of the add. Returns zero if not an iv.
2437 int PhaseIdealLoop::stride_of_possible_iv(Node* iff) {
2438 Node* trunc1 = NULL;
2439 Node* trunc2 = NULL;
2440 const TypeInteger* ttype = NULL;
2441 if (!iff->is_If() || iff->in(1) == NULL || !iff->in(1)->is_Bool()) {
2442 return 0;
2443 }
2444 BoolNode* bl = iff->in(1)->as_Bool();
2445 Node* cmp = bl->in(1);
2446 if (!cmp || (cmp->Opcode() != Op_CmpI && cmp->Opcode() != Op_CmpU)) {
2447 return 0;
2448 }
2449 // Must have an invariant operand
2450 if (is_member(get_loop(iff), get_ctrl(cmp->in(2)))) {
2451 return 0;
2452 }
2453 Node* add2 = NULL;
2454 Node* cmp1 = cmp->in(1);
2455 if (cmp1->is_Phi()) {
2456 // (If (Bool (CmpX phi:(Phi ...(Optional-trunc(AddI phi add2))) )))
2457 Node* phi = cmp1;
2458 for (uint i = 1; i < phi->req(); i++) {
2459 Node* in = phi->in(i);
2460 Node* add = CountedLoopNode::match_incr_with_optional_truncation(in,
2461 &trunc1, &trunc2, &ttype, T_INT);
2462 if (add && add->in(1) == phi) {
2463 add2 = add->in(2);
2464 break;
2465 }
2466 }
2467 } else {
2468 // (If (Bool (CmpX addtrunc:(Optional-trunc((AddI (Phi ...addtrunc...) add2)) )))
2469 Node* addtrunc = cmp1;
2470 Node* add = CountedLoopNode::match_incr_with_optional_truncation(addtrunc,
2471 &trunc1, &trunc2, &ttype, T_INT);
2472 if (add && add->in(1)->is_Phi()) {
2473 Node* phi = add->in(1);
2474 for (uint i = 1; i < phi->req(); i++) {
2475 if (phi->in(i) == addtrunc) {
2476 add2 = add->in(2);
2477 break;
2478 }
2479 }
2480 }
2481 }
2482 if (add2 != NULL) {
2483 const TypeInt* add2t = _igvn.type(add2)->is_int();
2484 if (add2t->is_con()) {
2485 return add2t->get_con();
2486 }
2487 }
2488 return 0;
2489 }
2490
2491
2492 //---------------------- stay_in_loop -------------------------------------
2493 // Return the (unique) control output node that's in the loop (if it exists.)
2494 Node* PhaseIdealLoop::stay_in_loop( Node* n, IdealLoopTree *loop) {
2495 Node* unique = NULL;
2496 if (!n) return NULL;
2497 for (DUIterator_Fast imax, i = n->fast_outs(imax); i < imax; i++) {
2498 Node* use = n->fast_out(i);
2499 if (!has_ctrl(use) && loop->is_member(get_loop(use))) {
2500 if (unique != NULL) {
2501 return NULL;
2502 }
2503 unique = use;
2504 }
2505 }
2506 return unique;
2507 }
2508
2509 //------------------------------ register_node -------------------------------------
2510 // Utility to register node "n" with PhaseIdealLoop
2511 void PhaseIdealLoop::register_node(Node* n, IdealLoopTree *loop, Node* pred, int ddepth) {
2512 _igvn.register_new_node_with_optimizer(n);
2513 loop->_body.push(n);
2514 if (n->is_CFG()) {
2515 set_loop(n, loop);
2516 set_idom(n, pred, ddepth);
2517 } else {
2518 set_ctrl(n, pred);
2519 }
2520 }
2521
2522 //------------------------------ proj_clone -------------------------------------
2523 // Utility to create an if-projection
2524 ProjNode* PhaseIdealLoop::proj_clone(ProjNode* p, IfNode* iff) {
2525 ProjNode* c = p->clone()->as_Proj();
2526 c->set_req(0, iff);
2527 return c;
2528 }
2529
2530 //------------------------------ short_circuit_if -------------------------------------
2531 // Force the iff control output to be the live_proj
2532 Node* PhaseIdealLoop::short_circuit_if(IfNode* iff, ProjNode* live_proj) {
2533 guarantee(live_proj != NULL, "null projection");
2534 int proj_con = live_proj->_con;
2535 assert(proj_con == 0 || proj_con == 1, "false or true projection");
2536 Node *con = _igvn.intcon(proj_con);
2537 set_ctrl(con, C->root());
2538 if (iff) {
2539 iff->set_req(1, con);
2540 }
2541 return con;
2542 }
2543
2544 //------------------------------ insert_if_before_proj -------------------------------------
2545 // Insert a new if before an if projection (* - new node)
2546 //
2547 // before
2548 // if(test)
2549 // / \
2550 // v v
2551 // other-proj proj (arg)
2552 //
2553 // after
2554 // if(test)
2555 // / \
2556 // / v
2557 // | * proj-clone
2558 // v |
2559 // other-proj v
2560 // * new_if(relop(cmp[IU](left,right)))
2561 // / \
2562 // v v
2563 // * new-proj proj
2564 // (returned)
2565 //
2566 ProjNode* PhaseIdealLoop::insert_if_before_proj(Node* left, bool Signed, BoolTest::mask relop, Node* right, ProjNode* proj) {
2567 IfNode* iff = proj->in(0)->as_If();
2568 IdealLoopTree *loop = get_loop(proj);
2569 ProjNode *other_proj = iff->proj_out(!proj->is_IfTrue())->as_Proj();
2570 int ddepth = dom_depth(proj);
2571
2572 _igvn.rehash_node_delayed(iff);
2573 _igvn.rehash_node_delayed(proj);
2574
2575 proj->set_req(0, NULL); // temporary disconnect
2576 ProjNode* proj2 = proj_clone(proj, iff);
2577 register_node(proj2, loop, iff, ddepth);
2578
2579 Node* cmp = Signed ? (Node*) new CmpINode(left, right) : (Node*) new CmpUNode(left, right);
2580 register_node(cmp, loop, proj2, ddepth);
2581
2582 BoolNode* bol = new BoolNode(cmp, relop);
2583 register_node(bol, loop, proj2, ddepth);
2584
2585 int opcode = iff->Opcode();
2586 assert(opcode == Op_If || opcode == Op_RangeCheck, "unexpected opcode");
2587 IfNode* new_if = (opcode == Op_If) ? new IfNode(proj2, bol, iff->_prob, iff->_fcnt):
2588 new RangeCheckNode(proj2, bol, iff->_prob, iff->_fcnt);
2589 register_node(new_if, loop, proj2, ddepth);
2590
2591 proj->set_req(0, new_if); // reattach
2592 set_idom(proj, new_if, ddepth);
2593
2594 ProjNode* new_exit = proj_clone(other_proj, new_if)->as_Proj();
2595 guarantee(new_exit != NULL, "null exit node");
2596 register_node(new_exit, get_loop(other_proj), new_if, ddepth);
2597
2598 return new_exit;
2599 }
2600
2601 //------------------------------ insert_region_before_proj -------------------------------------
2602 // Insert a region before an if projection (* - new node)
2603 //
2604 // before
2605 // if(test)
2606 // / |
2607 // v |
2608 // proj v
2609 // other-proj
2610 //
2611 // after
2612 // if(test)
2613 // / |
2614 // v |
2615 // * proj-clone v
2616 // | other-proj
2617 // v
2618 // * new-region
2619 // |
2620 // v
2621 // * dum_if
2622 // / \
2623 // v \
2624 // * dum-proj v
2625 // proj
2626 //
2627 RegionNode* PhaseIdealLoop::insert_region_before_proj(ProjNode* proj) {
2628 IfNode* iff = proj->in(0)->as_If();
2629 IdealLoopTree *loop = get_loop(proj);
2630 ProjNode *other_proj = iff->proj_out(!proj->is_IfTrue())->as_Proj();
2631 int ddepth = dom_depth(proj);
2632
2633 _igvn.rehash_node_delayed(iff);
2634 _igvn.rehash_node_delayed(proj);
2635
2636 proj->set_req(0, NULL); // temporary disconnect
2637 ProjNode* proj2 = proj_clone(proj, iff);
2638 register_node(proj2, loop, iff, ddepth);
2639
2640 RegionNode* reg = new RegionNode(2);
2641 reg->set_req(1, proj2);
2642 register_node(reg, loop, iff, ddepth);
2643
2644 IfNode* dum_if = new IfNode(reg, short_circuit_if(NULL, proj), iff->_prob, iff->_fcnt);
2645 register_node(dum_if, loop, reg, ddepth);
2646
2647 proj->set_req(0, dum_if); // reattach
2648 set_idom(proj, dum_if, ddepth);
2649
2650 ProjNode* dum_proj = proj_clone(other_proj, dum_if);
2651 register_node(dum_proj, loop, dum_if, ddepth);
2652
2653 return reg;
2654 }
2655
2656 //------------------------------ insert_cmpi_loop_exit -------------------------------------
2657 // Clone a signed compare loop exit from an unsigned compare and
2658 // insert it before the unsigned cmp on the stay-in-loop path.
2659 // All new nodes inserted in the dominator tree between the original
2660 // if and it's projections. The original if test is replaced with
2661 // a constant to force the stay-in-loop path.
2662 //
2663 // This is done to make sure that the original if and it's projections
2664 // still dominate the same set of control nodes, that the ctrl() relation
2665 // from data nodes to them is preserved, and that their loop nesting is
2666 // preserved.
2667 //
2668 // before
2669 // if(i <u limit) unsigned compare loop exit
2670 // / |
2671 // v v
2672 // exit-proj stay-in-loop-proj
2673 //
2674 // after
2675 // if(stay-in-loop-const) original if
2676 // / |
2677 // / v
2678 // / if(i < limit) new signed test
2679 // / / |
2680 // / / v
2681 // / / if(i <u limit) new cloned unsigned test
2682 // / / / |
2683 // v v v |
2684 // region |
2685 // | |
2686 // dum-if |
2687 // / | |
2688 // ether | |
2689 // v v
2690 // exit-proj stay-in-loop-proj
2691 //
2692 IfNode* PhaseIdealLoop::insert_cmpi_loop_exit(IfNode* if_cmpu, IdealLoopTree *loop) {
2693 const bool Signed = true;
2694 const bool Unsigned = false;
2695
2696 BoolNode* bol = if_cmpu->in(1)->as_Bool();
2697 if (bol->_test._test != BoolTest::lt) return NULL;
2698 CmpNode* cmpu = bol->in(1)->as_Cmp();
2699 if (cmpu->Opcode() != Op_CmpU) return NULL;
2700 int stride = stride_of_possible_iv(if_cmpu);
2701 if (stride == 0) return NULL;
2702
2703 Node* lp_proj = stay_in_loop(if_cmpu, loop);
2704 guarantee(lp_proj != NULL, "null loop node");
2705
2706 ProjNode* lp_continue = lp_proj->as_Proj();
2707 ProjNode* lp_exit = if_cmpu->proj_out(!lp_continue->is_IfTrue())->as_Proj();
2708 if (!lp_exit->is_IfFalse()) {
2709 // The loop exit condition is (i <u limit) ==> (i >= 0 && i < limit).
2710 // We therefore can't add a single exit condition.
2711 return NULL;
2712 }
2713 // The loop exit condition is !(i <u limit) ==> (i < 0 || i >= limit).
2714 // Split out the exit condition (i < 0) for stride < 0 or (i >= limit) for stride > 0.
2715 Node* limit = NULL;
2716 if (stride > 0) {
2717 limit = cmpu->in(2);
2718 } else {
2719 limit = _igvn.makecon(TypeInt::ZERO);
2720 set_ctrl(limit, C->root());
2721 }
2722 // Create a new region on the exit path
2723 RegionNode* reg = insert_region_before_proj(lp_exit);
2724 guarantee(reg != NULL, "null region node");
2725
2726 // Clone the if-cmpu-true-false using a signed compare
2727 BoolTest::mask rel_i = stride > 0 ? bol->_test._test : BoolTest::ge;
2728 ProjNode* cmpi_exit = insert_if_before_proj(cmpu->in(1), Signed, rel_i, limit, lp_continue);
2729 reg->add_req(cmpi_exit);
2730
2731 // Clone the if-cmpu-true-false
2732 BoolTest::mask rel_u = bol->_test._test;
2733 ProjNode* cmpu_exit = insert_if_before_proj(cmpu->in(1), Unsigned, rel_u, cmpu->in(2), lp_continue);
2734 reg->add_req(cmpu_exit);
2735
2736 // Force original if to stay in loop.
2737 short_circuit_if(if_cmpu, lp_continue);
2738
2739 return cmpi_exit->in(0)->as_If();
2740 }
2741
2742 //------------------------------ remove_cmpi_loop_exit -------------------------------------
2743 // Remove a previously inserted signed compare loop exit.
2744 void PhaseIdealLoop::remove_cmpi_loop_exit(IfNode* if_cmp, IdealLoopTree *loop) {
2745 Node* lp_proj = stay_in_loop(if_cmp, loop);
2746 assert(if_cmp->in(1)->in(1)->Opcode() == Op_CmpI &&
2747 stay_in_loop(lp_proj, loop)->is_If() &&
2748 stay_in_loop(lp_proj, loop)->in(1)->in(1)->Opcode() == Op_CmpU, "inserted cmpi before cmpu");
2749 Node *con = _igvn.makecon(lp_proj->is_IfTrue() ? TypeInt::ONE : TypeInt::ZERO);
2750 set_ctrl(con, C->root());
2751 if_cmp->set_req(1, con);
2752 }
2753
2754 //------------------------------ scheduled_nodelist -------------------------------------
2755 // Create a post order schedule of nodes that are in the
2756 // "member" set. The list is returned in "sched".
2757 // The first node in "sched" is the loop head, followed by
2758 // nodes which have no inputs in the "member" set, and then
2759 // followed by the nodes that have an immediate input dependence
2760 // on a node in "sched".
2761 void PhaseIdealLoop::scheduled_nodelist( IdealLoopTree *loop, VectorSet& member, Node_List &sched ) {
2762
2763 assert(member.test(loop->_head->_idx), "loop head must be in member set");
2764 VectorSet visited;
2765 Node_Stack nstack(loop->_body.size());
2766
2767 Node* n = loop->_head; // top of stack is cached in "n"
2768 uint idx = 0;
2769 visited.set(n->_idx);
2770
2771 // Initially push all with no inputs from within member set
2772 for(uint i = 0; i < loop->_body.size(); i++ ) {
2773 Node *elt = loop->_body.at(i);
2774 if (member.test(elt->_idx)) {
2775 bool found = false;
2776 for (uint j = 0; j < elt->req(); j++) {
2777 Node* def = elt->in(j);
2778 if (def && member.test(def->_idx) && def != elt) {
2779 found = true;
2780 break;
2781 }
2782 }
2783 if (!found && elt != loop->_head) {
2784 nstack.push(n, idx);
2785 n = elt;
2786 assert(!visited.test(n->_idx), "not seen yet");
2787 visited.set(n->_idx);
2788 }
2789 }
2790 }
2791
2792 // traverse out's that are in the member set
2793 while (true) {
2794 if (idx < n->outcnt()) {
2795 Node* use = n->raw_out(idx);
2796 idx++;
2797 if (!visited.test_set(use->_idx)) {
2798 if (member.test(use->_idx)) {
2799 nstack.push(n, idx);
2800 n = use;
2801 idx = 0;
2802 }
2803 }
2804 } else {
2805 // All outputs processed
2806 sched.push(n);
2807 if (nstack.is_empty()) break;
2808 n = nstack.node();
2809 idx = nstack.index();
2810 nstack.pop();
2811 }
2812 }
2813 }
2814
2815
2816 //------------------------------ has_use_in_set -------------------------------------
2817 // Has a use in the vector set
2818 bool PhaseIdealLoop::has_use_in_set( Node* n, VectorSet& vset ) {
2819 for (DUIterator_Fast jmax, j = n->fast_outs(jmax); j < jmax; j++) {
2820 Node* use = n->fast_out(j);
2821 if (vset.test(use->_idx)) {
2822 return true;
2823 }
2824 }
2825 return false;
2826 }
2827
2828
2829 //------------------------------ has_use_internal_to_set -------------------------------------
2830 // Has use internal to the vector set (ie. not in a phi at the loop head)
2831 bool PhaseIdealLoop::has_use_internal_to_set( Node* n, VectorSet& vset, IdealLoopTree *loop ) {
2832 Node* head = loop->_head;
2833 for (DUIterator_Fast jmax, j = n->fast_outs(jmax); j < jmax; j++) {
2834 Node* use = n->fast_out(j);
2835 if (vset.test(use->_idx) && !(use->is_Phi() && use->in(0) == head)) {
2836 return true;
2837 }
2838 }
2839 return false;
2840 }
2841
2842
2843 //------------------------------ clone_for_use_outside_loop -------------------------------------
2844 // clone "n" for uses that are outside of loop
2845 int PhaseIdealLoop::clone_for_use_outside_loop( IdealLoopTree *loop, Node* n, Node_List& worklist ) {
2846 int cloned = 0;
2847 assert(worklist.size() == 0, "should be empty");
2848 for (DUIterator_Fast jmax, j = n->fast_outs(jmax); j < jmax; j++) {
2849 Node* use = n->fast_out(j);
2850 if( !loop->is_member(get_loop(has_ctrl(use) ? get_ctrl(use) : use)) ) {
2851 worklist.push(use);
2852 }
2853 }
2854
2855 if (C->check_node_count(worklist.size() + NodeLimitFudgeFactor,
2856 "Too many clones required in clone_for_use_outside_loop in partial peeling")) {
2857 return -1;
2858 }
2859
2860 while( worklist.size() ) {
2861 Node *use = worklist.pop();
2862 if (!has_node(use) || use->in(0) == C->top()) continue;
2863 uint j;
2864 for (j = 0; j < use->req(); j++) {
2865 if (use->in(j) == n) break;
2866 }
2867 assert(j < use->req(), "must be there");
2868
2869 // clone "n" and insert it between the inputs of "n" and the use outside the loop
2870 Node* n_clone = n->clone();
2871 _igvn.replace_input_of(use, j, n_clone);
2872 cloned++;
2873 Node* use_c;
2874 if (!use->is_Phi()) {
2875 use_c = has_ctrl(use) ? get_ctrl(use) : use->in(0);
2876 } else {
2877 // Use in a phi is considered a use in the associated predecessor block
2878 use_c = use->in(0)->in(j);
2879 }
2880 set_ctrl(n_clone, use_c);
2881 assert(!loop->is_member(get_loop(use_c)), "should be outside loop");
2882 get_loop(use_c)->_body.push(n_clone);
2883 _igvn.register_new_node_with_optimizer(n_clone);
2884 #ifndef PRODUCT
2885 if (TracePartialPeeling) {
2886 tty->print_cr("loop exit cloning old: %d new: %d newbb: %d", n->_idx, n_clone->_idx, get_ctrl(n_clone)->_idx);
2887 }
2888 #endif
2889 }
2890 return cloned;
2891 }
2892
2893
2894 //------------------------------ clone_for_special_use_inside_loop -------------------------------------
2895 // clone "n" for special uses that are in the not_peeled region.
2896 // If these def-uses occur in separate blocks, the code generator
2897 // marks the method as not compilable. For example, if a "BoolNode"
2898 // is in a different basic block than the "IfNode" that uses it, then
2899 // the compilation is aborted in the code generator.
2900 void PhaseIdealLoop::clone_for_special_use_inside_loop( IdealLoopTree *loop, Node* n,
2901 VectorSet& not_peel, Node_List& sink_list, Node_List& worklist ) {
2902 if (n->is_Phi() || n->is_Load()) {
2903 return;
2904 }
2905 assert(worklist.size() == 0, "should be empty");
2906 for (DUIterator_Fast jmax, j = n->fast_outs(jmax); j < jmax; j++) {
2907 Node* use = n->fast_out(j);
2908 if ( not_peel.test(use->_idx) &&
2909 (use->is_If() || use->is_CMove() || use->is_Bool()) &&
2910 use->in(1) == n) {
2911 worklist.push(use);
2912 }
2913 }
2914 if (worklist.size() > 0) {
2915 // clone "n" and insert it between inputs of "n" and the use
2916 Node* n_clone = n->clone();
2917 loop->_body.push(n_clone);
2918 _igvn.register_new_node_with_optimizer(n_clone);
2919 set_ctrl(n_clone, get_ctrl(n));
2920 sink_list.push(n_clone);
2921 not_peel.set(n_clone->_idx);
2922 #ifndef PRODUCT
2923 if (TracePartialPeeling) {
2924 tty->print_cr("special not_peeled cloning old: %d new: %d", n->_idx, n_clone->_idx);
2925 }
2926 #endif
2927 while( worklist.size() ) {
2928 Node *use = worklist.pop();
2929 _igvn.rehash_node_delayed(use);
2930 for (uint j = 1; j < use->req(); j++) {
2931 if (use->in(j) == n) {
2932 use->set_req(j, n_clone);
2933 }
2934 }
2935 }
2936 }
2937 }
2938
2939
2940 //------------------------------ insert_phi_for_loop -------------------------------------
2941 // Insert phi(lp_entry_val, back_edge_val) at use->in(idx) for loop lp if phi does not already exist
2942 void PhaseIdealLoop::insert_phi_for_loop( Node* use, uint idx, Node* lp_entry_val, Node* back_edge_val, LoopNode* lp ) {
2943 Node *phi = PhiNode::make(lp, back_edge_val);
2944 phi->set_req(LoopNode::EntryControl, lp_entry_val);
2945 // Use existing phi if it already exists
2946 Node *hit = _igvn.hash_find_insert(phi);
2947 if( hit == NULL ) {
2948 _igvn.register_new_node_with_optimizer(phi);
2949 set_ctrl(phi, lp);
2950 } else {
2951 // Remove the new phi from the graph and use the hit
2952 _igvn.remove_dead_node(phi);
2953 phi = hit;
2954 }
2955 _igvn.replace_input_of(use, idx, phi);
2956 }
2957
2958 #ifdef ASSERT
2959 //------------------------------ is_valid_loop_partition -------------------------------------
2960 // Validate the loop partition sets: peel and not_peel
2961 bool PhaseIdealLoop::is_valid_loop_partition( IdealLoopTree *loop, VectorSet& peel, Node_List& peel_list,
2962 VectorSet& not_peel ) {
2963 uint i;
2964 // Check that peel_list entries are in the peel set
2965 for (i = 0; i < peel_list.size(); i++) {
2966 if (!peel.test(peel_list.at(i)->_idx)) {
2967 return false;
2968 }
2969 }
2970 // Check at loop members are in one of peel set or not_peel set
2971 for (i = 0; i < loop->_body.size(); i++ ) {
2972 Node *def = loop->_body.at(i);
2973 uint di = def->_idx;
2974 // Check that peel set elements are in peel_list
2975 if (peel.test(di)) {
2976 if (not_peel.test(di)) {
2977 return false;
2978 }
2979 // Must be in peel_list also
2980 bool found = false;
2981 for (uint j = 0; j < peel_list.size(); j++) {
2982 if (peel_list.at(j)->_idx == di) {
2983 found = true;
2984 break;
2985 }
2986 }
2987 if (!found) {
2988 return false;
2989 }
2990 } else if (not_peel.test(di)) {
2991 if (peel.test(di)) {
2992 return false;
2993 }
2994 } else {
2995 return false;
2996 }
2997 }
2998 return true;
2999 }
3000
3001 //------------------------------ is_valid_clone_loop_exit_use -------------------------------------
3002 // Ensure a use outside of loop is of the right form
3003 bool PhaseIdealLoop::is_valid_clone_loop_exit_use( IdealLoopTree *loop, Node* use, uint exit_idx) {
3004 Node *use_c = has_ctrl(use) ? get_ctrl(use) : use;
3005 return (use->is_Phi() &&
3006 use_c->is_Region() && use_c->req() == 3 &&
3007 (use_c->in(exit_idx)->Opcode() == Op_IfTrue ||
3008 use_c->in(exit_idx)->Opcode() == Op_IfFalse ||
3009 use_c->in(exit_idx)->Opcode() == Op_JumpProj) &&
3010 loop->is_member( get_loop( use_c->in(exit_idx)->in(0) ) ) );
3011 }
3012
3013 //------------------------------ is_valid_clone_loop_form -------------------------------------
3014 // Ensure that all uses outside of loop are of the right form
3015 bool PhaseIdealLoop::is_valid_clone_loop_form( IdealLoopTree *loop, Node_List& peel_list,
3016 uint orig_exit_idx, uint clone_exit_idx) {
3017 uint len = peel_list.size();
3018 for (uint i = 0; i < len; i++) {
3019 Node *def = peel_list.at(i);
3020
3021 for (DUIterator_Fast jmax, j = def->fast_outs(jmax); j < jmax; j++) {
3022 Node *use = def->fast_out(j);
3023 Node *use_c = has_ctrl(use) ? get_ctrl(use) : use;
3024 if (!loop->is_member(get_loop(use_c))) {
3025 // use is not in the loop, check for correct structure
3026 if (use->in(0) == def) {
3027 // Okay
3028 } else if (!is_valid_clone_loop_exit_use(loop, use, orig_exit_idx)) {
3029 return false;
3030 }
3031 }
3032 }
3033 }
3034 return true;
3035 }
3036 #endif
3037
3038 //------------------------------ partial_peel -------------------------------------
3039 // Partially peel (aka loop rotation) the top portion of a loop (called
3040 // the peel section below) by cloning it and placing one copy just before
3041 // the new loop head and the other copy at the bottom of the new loop.
3042 //
3043 // before after where it came from
3044 //
3045 // stmt1 stmt1
3046 // loop: stmt2 clone
3047 // stmt2 if condA goto exitA clone
3048 // if condA goto exitA new_loop: new
3049 // stmt3 stmt3 clone
3050 // if !condB goto loop if condB goto exitB clone
3051 // exitB: stmt2 orig
3052 // stmt4 if !condA goto new_loop orig
3053 // exitA: goto exitA
3054 // exitB:
3055 // stmt4
3056 // exitA:
3057 //
3058 // Step 1: find the cut point: an exit test on probable
3059 // induction variable.
3060 // Step 2: schedule (with cloning) operations in the peel
3061 // section that can be executed after the cut into
3062 // the section that is not peeled. This may need
3063 // to clone operations into exit blocks. For
3064 // instance, a reference to A[i] in the not-peel
3065 // section and a reference to B[i] in an exit block
3066 // may cause a left-shift of i by 2 to be placed
3067 // in the peel block. This step will clone the left
3068 // shift into the exit block and sink the left shift
3069 // from the peel to the not-peel section.
3070 // Step 3: clone the loop, retarget the control, and insert
3071 // phis for values that are live across the new loop
3072 // head. This is very dependent on the graph structure
3073 // from clone_loop. It creates region nodes for
3074 // exit control and associated phi nodes for values
3075 // flow out of the loop through that exit. The region
3076 // node is dominated by the clone's control projection.
3077 // So the clone's peel section is placed before the
3078 // new loop head, and the clone's not-peel section is
3079 // forms the top part of the new loop. The original
3080 // peel section forms the tail of the new loop.
3081 // Step 4: update the dominator tree and recompute the
3082 // dominator depth.
3083 //
3084 // orig
3085 //
3086 // stmt1
3087 // |
3088 // v
3089 // loop predicate
3090 // |
3091 // v
3092 // loop<----+
3093 // | |
3094 // stmt2 |
3095 // | |
3096 // v |
3097 // ifA |
3098 // / | |
3099 // v v |
3100 // false true ^ <-- last_peel
3101 // / | |
3102 // / ===|==cut |
3103 // / stmt3 | <-- first_not_peel
3104 // / | |
3105 // | v |
3106 // v ifB |
3107 // exitA: / \ |
3108 // / \ |
3109 // v v |
3110 // false true |
3111 // / \ |
3112 // / ----+
3113 // |
3114 // v
3115 // exitB:
3116 // stmt4
3117 //
3118 //
3119 // after clone loop
3120 //
3121 // stmt1
3122 // |
3123 // v
3124 // loop predicate
3125 // / \
3126 // clone / \ orig
3127 // / \
3128 // / \
3129 // v v
3130 // +---->loop loop<----+
3131 // | | | |
3132 // | stmt2 stmt2 |
3133 // | | | |
3134 // | v v |
3135 // | ifA ifA |
3136 // | | \ / | |
3137 // | v v v v |
3138 // ^ true false false true ^ <-- last_peel
3139 // | | ^ \ / | |
3140 // | cut==|== \ \ / ===|==cut |
3141 // | stmt3 \ \ / stmt3 | <-- first_not_peel
3142 // | | dom | | | |
3143 // | v \ 1v v2 v |
3144 // | ifB regionA ifB |
3145 // | / \ | / \ |
3146 // | / \ v / \ |
3147 // | v v exitA: v v |
3148 // | true false false true |
3149 // | / ^ \ / \ |
3150 // +---- \ \ / ----+
3151 // dom \ /
3152 // \ 1v v2
3153 // regionB
3154 // |
3155 // v
3156 // exitB:
3157 // stmt4
3158 //
3159 //
3160 // after partial peel
3161 //
3162 // stmt1
3163 // |
3164 // v
3165 // loop predicate
3166 // /
3167 // clone / orig
3168 // / TOP
3169 // / \
3170 // v v
3171 // TOP->loop loop----+
3172 // | | |
3173 // stmt2 stmt2 |
3174 // | | |
3175 // v v |
3176 // ifA ifA |
3177 // | \ / | |
3178 // v v v v |
3179 // true false false true | <-- last_peel
3180 // | ^ \ / +------|---+
3181 // +->newloop \ \ / === ==cut | |
3182 // | stmt3 \ \ / TOP | |
3183 // | | dom | | stmt3 | | <-- first_not_peel
3184 // | v \ 1v v2 v | |
3185 // | ifB regionA ifB ^ v
3186 // | / \ | / \ | |
3187 // | / \ v / \ | |
3188 // | v v exitA: v v | |
3189 // | true false false true | |
3190 // | / ^ \ / \ | |
3191 // | | \ \ / v | |
3192 // | | dom \ / TOP | |
3193 // | | \ 1v v2 | |
3194 // ^ v regionB | |
3195 // | | | | |
3196 // | | v ^ v
3197 // | | exitB: | |
3198 // | | stmt4 | |
3199 // | +------------>-----------------+ |
3200 // | |
3201 // +-----------------<---------------------+
3202 //
3203 //
3204 // final graph
3205 //
3206 // stmt1
3207 // |
3208 // v
3209 // loop predicate
3210 // |
3211 // v
3212 // stmt2 clone
3213 // |
3214 // v
3215 // ........> ifA clone
3216 // : / |
3217 // dom / |
3218 // : v v
3219 // : false true
3220 // : | |
3221 // : | v
3222 // : | newloop<-----+
3223 // : | | |
3224 // : | stmt3 clone |
3225 // : | | |
3226 // : | v |
3227 // : | ifB |
3228 // : | / \ |
3229 // : | v v |
3230 // : | false true |
3231 // : | | | |
3232 // : | v stmt2 |
3233 // : | exitB: | |
3234 // : | stmt4 v |
3235 // : | ifA orig |
3236 // : | / \ |
3237 // : | / \ |
3238 // : | v v |
3239 // : | false true |
3240 // : | / \ |
3241 // : v v -----+
3242 // RegionA
3243 // |
3244 // v
3245 // exitA
3246 //
3247 bool PhaseIdealLoop::partial_peel( IdealLoopTree *loop, Node_List &old_new ) {
3248
3249 assert(!loop->_head->is_CountedLoop(), "Non-counted loop only");
3250 if (!loop->_head->is_Loop()) {
3251 return false;
3252 }
3253 LoopNode *head = loop->_head->as_Loop();
3254
3255 if (head->is_partial_peel_loop() || head->partial_peel_has_failed()) {
3256 return false;
3257 }
3258
3259 // Check for complex exit control
3260 for (uint ii = 0; ii < loop->_body.size(); ii++) {
3261 Node *n = loop->_body.at(ii);
3262 int opc = n->Opcode();
3263 if (n->is_Call() ||
3264 opc == Op_Catch ||
3265 opc == Op_CatchProj ||
3266 opc == Op_Jump ||
3267 opc == Op_JumpProj) {
3268 #ifndef PRODUCT
3269 if (TracePartialPeeling) {
3270 tty->print_cr("\nExit control too complex: lp: %d", head->_idx);
3271 }
3272 #endif
3273 return false;
3274 }
3275 }
3276
3277 int dd = dom_depth(head);
3278
3279 // Step 1: find cut point
3280
3281 // Walk up dominators to loop head looking for first loop exit
3282 // which is executed on every path thru loop.
3283 IfNode *peel_if = NULL;
3284 IfNode *peel_if_cmpu = NULL;
3285
3286 Node *iff = loop->tail();
3287 while (iff != head) {
3288 if (iff->is_If()) {
3289 Node *ctrl = get_ctrl(iff->in(1));
3290 if (ctrl->is_top()) return false; // Dead test on live IF.
3291 // If loop-varying exit-test, check for induction variable
3292 if (loop->is_member(get_loop(ctrl)) &&
3293 loop->is_loop_exit(iff) &&
3294 is_possible_iv_test(iff)) {
3295 Node* cmp = iff->in(1)->in(1);
3296 if (cmp->Opcode() == Op_CmpI) {
3297 peel_if = iff->as_If();
3298 } else {
3299 assert(cmp->Opcode() == Op_CmpU, "must be CmpI or CmpU");
3300 peel_if_cmpu = iff->as_If();
3301 }
3302 }
3303 }
3304 iff = idom(iff);
3305 }
3306
3307 // Prefer signed compare over unsigned compare.
3308 IfNode* new_peel_if = NULL;
3309 if (peel_if == NULL) {
3310 if (!PartialPeelAtUnsignedTests || peel_if_cmpu == NULL) {
3311 return false; // No peel point found
3312 }
3313 new_peel_if = insert_cmpi_loop_exit(peel_if_cmpu, loop);
3314 if (new_peel_if == NULL) {
3315 return false; // No peel point found
3316 }
3317 peel_if = new_peel_if;
3318 }
3319 Node* last_peel = stay_in_loop(peel_if, loop);
3320 Node* first_not_peeled = stay_in_loop(last_peel, loop);
3321 if (first_not_peeled == NULL || first_not_peeled == head) {
3322 return false;
3323 }
3324
3325 #ifndef PRODUCT
3326 if (TraceLoopOpts) {
3327 tty->print("PartialPeel ");
3328 loop->dump_head();
3329 }
3330
3331 if (TracePartialPeeling) {
3332 tty->print_cr("before partial peel one iteration");
3333 Node_List wl;
3334 Node* t = head->in(2);
3335 while (true) {
3336 wl.push(t);
3337 if (t == head) break;
3338 t = idom(t);
3339 }
3340 while (wl.size() > 0) {
3341 Node* tt = wl.pop();
3342 tt->dump();
3343 if (tt == last_peel) tty->print_cr("-- cut --");
3344 }
3345 }
3346 #endif
3347 VectorSet peel;
3348 VectorSet not_peel;
3349 Node_List peel_list;
3350 Node_List worklist;
3351 Node_List sink_list;
3352
3353 uint estimate = loop->est_loop_clone_sz(1);
3354 if (exceeding_node_budget(estimate)) {
3355 return false;
3356 }
3357
3358 // Set of cfg nodes to peel are those that are executable from
3359 // the head through last_peel.
3360 assert(worklist.size() == 0, "should be empty");
3361 worklist.push(head);
3362 peel.set(head->_idx);
3363 while (worklist.size() > 0) {
3364 Node *n = worklist.pop();
3365 if (n != last_peel) {
3366 for (DUIterator_Fast jmax, j = n->fast_outs(jmax); j < jmax; j++) {
3367 Node* use = n->fast_out(j);
3368 if (use->is_CFG() &&
3369 loop->is_member(get_loop(use)) &&
3370 !peel.test_set(use->_idx)) {
3371 worklist.push(use);
3372 }
3373 }
3374 }
3375 }
3376
3377 // Set of non-cfg nodes to peel are those that are control
3378 // dependent on the cfg nodes.
3379 for (uint i = 0; i < loop->_body.size(); i++) {
3380 Node *n = loop->_body.at(i);
3381 Node *n_c = has_ctrl(n) ? get_ctrl(n) : n;
3382 if (peel.test(n_c->_idx)) {
3383 peel.set(n->_idx);
3384 } else {
3385 not_peel.set(n->_idx);
3386 }
3387 }
3388
3389 // Step 2: move operations from the peeled section down into the
3390 // not-peeled section
3391
3392 // Get a post order schedule of nodes in the peel region
3393 // Result in right-most operand.
3394 scheduled_nodelist(loop, peel, peel_list);
3395
3396 assert(is_valid_loop_partition(loop, peel, peel_list, not_peel), "bad partition");
3397
3398 // For future check for too many new phis
3399 uint old_phi_cnt = 0;
3400 for (DUIterator_Fast jmax, j = head->fast_outs(jmax); j < jmax; j++) {
3401 Node* use = head->fast_out(j);
3402 if (use->is_Phi()) old_phi_cnt++;
3403 }
3404
3405 #ifndef PRODUCT
3406 if (TracePartialPeeling) {
3407 tty->print_cr("\npeeled list");
3408 }
3409 #endif
3410
3411 // Evacuate nodes in peel region into the not_peeled region if possible
3412 bool too_many_clones = false;
3413 uint new_phi_cnt = 0;
3414 uint cloned_for_outside_use = 0;
3415 for (uint i = 0; i < peel_list.size();) {
3416 Node* n = peel_list.at(i);
3417 #ifndef PRODUCT
3418 if (TracePartialPeeling) n->dump();
3419 #endif
3420 bool incr = true;
3421 if (!n->is_CFG()) {
3422 if (has_use_in_set(n, not_peel)) {
3423 // If not used internal to the peeled region,
3424 // move "n" from peeled to not_peeled region.
3425 if (!has_use_internal_to_set(n, peel, loop)) {
3426 // if not pinned and not a load (which maybe anti-dependent on a store)
3427 // and not a CMove (Matcher expects only bool->cmove).
3428 if (n->in(0) == NULL && !n->is_Load() && !n->is_CMove()) {
3429 int new_clones = clone_for_use_outside_loop(loop, n, worklist);
3430 if (new_clones == -1) {
3431 too_many_clones = true;
3432 break;
3433 }
3434 cloned_for_outside_use += new_clones;
3435 sink_list.push(n);
3436 peel.remove(n->_idx);
3437 not_peel.set(n->_idx);
3438 peel_list.remove(i);
3439 incr = false;
3440 #ifndef PRODUCT
3441 if (TracePartialPeeling) {
3442 tty->print_cr("sink to not_peeled region: %d newbb: %d",
3443 n->_idx, get_ctrl(n)->_idx);
3444 }
3445 #endif
3446 }
3447 } else {
3448 // Otherwise check for special def-use cases that span
3449 // the peel/not_peel boundary such as bool->if
3450 clone_for_special_use_inside_loop(loop, n, not_peel, sink_list, worklist);
3451 new_phi_cnt++;
3452 }
3453 }
3454 }
3455 if (incr) i++;
3456 }
3457
3458 estimate += cloned_for_outside_use + new_phi_cnt;
3459 bool exceed_node_budget = !may_require_nodes(estimate);
3460 bool exceed_phi_limit = new_phi_cnt > old_phi_cnt + PartialPeelNewPhiDelta;
3461
3462 if (too_many_clones || exceed_node_budget || exceed_phi_limit) {
3463 #ifndef PRODUCT
3464 if (TracePartialPeeling && exceed_phi_limit) {
3465 tty->print_cr("\nToo many new phis: %d old %d new cmpi: %c",
3466 new_phi_cnt, old_phi_cnt, new_peel_if != NULL?'T':'F');
3467 }
3468 #endif
3469 if (new_peel_if != NULL) {
3470 remove_cmpi_loop_exit(new_peel_if, loop);
3471 }
3472 // Inhibit more partial peeling on this loop
3473 assert(!head->is_partial_peel_loop(), "not partial peeled");
3474 head->mark_partial_peel_failed();
3475 if (cloned_for_outside_use > 0) {
3476 // Terminate this round of loop opts because
3477 // the graph outside this loop was changed.
3478 C->set_major_progress();
3479 return true;
3480 }
3481 return false;
3482 }
3483
3484 // Step 3: clone loop, retarget control, and insert new phis
3485
3486 // Create new loop head for new phis and to hang
3487 // the nodes being moved (sinked) from the peel region.
3488 LoopNode* new_head = new LoopNode(last_peel, last_peel);
3489 new_head->set_unswitch_count(head->unswitch_count()); // Preserve
3490 _igvn.register_new_node_with_optimizer(new_head);
3491 assert(first_not_peeled->in(0) == last_peel, "last_peel <- first_not_peeled");
3492 _igvn.replace_input_of(first_not_peeled, 0, new_head);
3493 set_loop(new_head, loop);
3494 loop->_body.push(new_head);
3495 not_peel.set(new_head->_idx);
3496 set_idom(new_head, last_peel, dom_depth(first_not_peeled));
3497 set_idom(first_not_peeled, new_head, dom_depth(first_not_peeled));
3498
3499 while (sink_list.size() > 0) {
3500 Node* n = sink_list.pop();
3501 set_ctrl(n, new_head);
3502 }
3503
3504 assert(is_valid_loop_partition(loop, peel, peel_list, not_peel), "bad partition");
3505
3506 clone_loop(loop, old_new, dd, IgnoreStripMined);
3507
3508 const uint clone_exit_idx = 1;
3509 const uint orig_exit_idx = 2;
3510 assert(is_valid_clone_loop_form(loop, peel_list, orig_exit_idx, clone_exit_idx), "bad clone loop");
3511
3512 Node* head_clone = old_new[head->_idx];
3513 LoopNode* new_head_clone = old_new[new_head->_idx]->as_Loop();
3514 Node* orig_tail_clone = head_clone->in(2);
3515
3516 // Add phi if "def" node is in peel set and "use" is not
3517
3518 for (uint i = 0; i < peel_list.size(); i++) {
3519 Node *def = peel_list.at(i);
3520 if (!def->is_CFG()) {
3521 for (DUIterator_Fast jmax, j = def->fast_outs(jmax); j < jmax; j++) {
3522 Node *use = def->fast_out(j);
3523 if (has_node(use) && use->in(0) != C->top() &&
3524 (!peel.test(use->_idx) ||
3525 (use->is_Phi() && use->in(0) == head)) ) {
3526 worklist.push(use);
3527 }
3528 }
3529 while( worklist.size() ) {
3530 Node *use = worklist.pop();
3531 for (uint j = 1; j < use->req(); j++) {
3532 Node* n = use->in(j);
3533 if (n == def) {
3534
3535 // "def" is in peel set, "use" is not in peel set
3536 // or "use" is in the entry boundary (a phi) of the peel set
3537
3538 Node* use_c = has_ctrl(use) ? get_ctrl(use) : use;
3539
3540 if ( loop->is_member(get_loop( use_c )) ) {
3541 // use is in loop
3542 if (old_new[use->_idx] != NULL) { // null for dead code
3543 Node* use_clone = old_new[use->_idx];
3544 _igvn.replace_input_of(use, j, C->top());
3545 insert_phi_for_loop( use_clone, j, old_new[def->_idx], def, new_head_clone );
3546 }
3547 } else {
3548 assert(is_valid_clone_loop_exit_use(loop, use, orig_exit_idx), "clone loop format");
3549 // use is not in the loop, check if the live range includes the cut
3550 Node* lp_if = use_c->in(orig_exit_idx)->in(0);
3551 if (not_peel.test(lp_if->_idx)) {
3552 assert(j == orig_exit_idx, "use from original loop");
3553 insert_phi_for_loop( use, clone_exit_idx, old_new[def->_idx], def, new_head_clone );
3554 }
3555 }
3556 }
3557 }
3558 }
3559 }
3560 }
3561
3562 // Step 3b: retarget control
3563
3564 // Redirect control to the new loop head if a cloned node in
3565 // the not_peeled region has control that points into the peeled region.
3566 // This necessary because the cloned peeled region will be outside
3567 // the loop.
3568 // from to
3569 // cloned-peeled <---+
3570 // new_head_clone: | <--+
3571 // cloned-not_peeled in(0) in(0)
3572 // orig-peeled
3573
3574 for (uint i = 0; i < loop->_body.size(); i++) {
3575 Node *n = loop->_body.at(i);
3576 if (!n->is_CFG() && n->in(0) != NULL &&
3577 not_peel.test(n->_idx) && peel.test(n->in(0)->_idx)) {
3578 Node* n_clone = old_new[n->_idx];
3579 _igvn.replace_input_of(n_clone, 0, new_head_clone);
3580 }
3581 }
3582
3583 // Backedge of the surviving new_head (the clone) is original last_peel
3584 _igvn.replace_input_of(new_head_clone, LoopNode::LoopBackControl, last_peel);
3585
3586 // Cut first node in original not_peel set
3587 _igvn.rehash_node_delayed(new_head); // Multiple edge updates:
3588 new_head->set_req(LoopNode::EntryControl, C->top()); // use rehash_node_delayed / set_req instead of
3589 new_head->set_req(LoopNode::LoopBackControl, C->top()); // multiple replace_input_of calls
3590
3591 // Copy head_clone back-branch info to original head
3592 // and remove original head's loop entry and
3593 // clone head's back-branch
3594 _igvn.rehash_node_delayed(head); // Multiple edge updates
3595 head->set_req(LoopNode::EntryControl, head_clone->in(LoopNode::LoopBackControl));
3596 head->set_req(LoopNode::LoopBackControl, C->top());
3597 _igvn.replace_input_of(head_clone, LoopNode::LoopBackControl, C->top());
3598
3599 // Similarly modify the phis
3600 for (DUIterator_Fast kmax, k = head->fast_outs(kmax); k < kmax; k++) {
3601 Node* use = head->fast_out(k);
3602 if (use->is_Phi() && use->outcnt() > 0) {
3603 Node* use_clone = old_new[use->_idx];
3604 _igvn.rehash_node_delayed(use); // Multiple edge updates
3605 use->set_req(LoopNode::EntryControl, use_clone->in(LoopNode::LoopBackControl));
3606 use->set_req(LoopNode::LoopBackControl, C->top());
3607 _igvn.replace_input_of(use_clone, LoopNode::LoopBackControl, C->top());
3608 }
3609 }
3610
3611 // Step 4: update dominator tree and dominator depth
3612
3613 set_idom(head, orig_tail_clone, dd);
3614 recompute_dom_depth();
3615
3616 // Inhibit more partial peeling on this loop
3617 new_head_clone->set_partial_peel_loop();
3618 C->set_major_progress();
3619 loop->record_for_igvn();
3620
3621 #ifndef PRODUCT
3622 if (TracePartialPeeling) {
3623 tty->print_cr("\nafter partial peel one iteration");
3624 Node_List wl;
3625 Node* t = last_peel;
3626 while (true) {
3627 wl.push(t);
3628 if (t == head_clone) break;
3629 t = idom(t);
3630 }
3631 while (wl.size() > 0) {
3632 Node* tt = wl.pop();
3633 if (tt == head) tty->print_cr("orig head");
3634 else if (tt == new_head_clone) tty->print_cr("new head");
3635 else if (tt == head_clone) tty->print_cr("clone head");
3636 tt->dump();
3637 }
3638 }
3639 #endif
3640 return true;
3641 }
3642
3643 //------------------------------reorg_offsets----------------------------------
3644 // Reorganize offset computations to lower register pressure. Mostly
3645 // prevent loop-fallout uses of the pre-incremented trip counter (which are
3646 // then alive with the post-incremented trip counter forcing an extra
3647 // register move)
3648 void PhaseIdealLoop::reorg_offsets(IdealLoopTree *loop) {
3649 // Perform it only for canonical counted loops.
3650 // Loop's shape could be messed up by iteration_split_impl.
3651 if (!loop->_head->is_CountedLoop())
3652 return;
3653 if (!loop->_head->as_Loop()->is_valid_counted_loop(T_INT))
3654 return;
3655
3656 CountedLoopNode *cl = loop->_head->as_CountedLoop();
3657 CountedLoopEndNode *cle = cl->loopexit();
3658 Node *exit = cle->proj_out(false);
3659 Node *phi = cl->phi();
3660
3661 // Check for the special case when using the pre-incremented trip-counter on
3662 // the fall-out path (forces the pre-incremented and post-incremented trip
3663 // counter to be live at the same time). Fix this by adjusting to use the
3664 // post-increment trip counter.
3665
3666 bool progress = true;
3667 while (progress) {
3668 progress = false;
3669 for (DUIterator_Fast imax, i = phi->fast_outs(imax); i < imax; i++) {
3670 Node* use = phi->fast_out(i); // User of trip-counter
3671 if (!has_ctrl(use)) continue;
3672 Node *u_ctrl = get_ctrl(use);
3673 if (use->is_Phi()) {
3674 u_ctrl = NULL;
3675 for (uint j = 1; j < use->req(); j++)
3676 if (use->in(j) == phi)
3677 u_ctrl = dom_lca(u_ctrl, use->in(0)->in(j));
3678 }
3679 IdealLoopTree *u_loop = get_loop(u_ctrl);
3680 // Look for loop-invariant use
3681 if (u_loop == loop) continue;
3682 if (loop->is_member(u_loop)) continue;
3683 // Check that use is live out the bottom. Assuming the trip-counter
3684 // update is right at the bottom, uses of of the loop middle are ok.
3685 if (dom_lca(exit, u_ctrl) != exit) continue;
3686 // Hit! Refactor use to use the post-incremented tripcounter.
3687 // Compute a post-increment tripcounter.
3688 Node* c = exit;
3689 if (cl->is_strip_mined()) {
3690 IdealLoopTree* outer_loop = get_loop(cl->outer_loop());
3691 if (!outer_loop->is_member(u_loop)) {
3692 c = cl->outer_loop_exit();
3693 }
3694 }
3695 Node *opaq = new Opaque2Node(C, cle->incr());
3696 register_new_node(opaq, c);
3697 Node *neg_stride = _igvn.intcon(-cle->stride_con());
3698 set_ctrl(neg_stride, C->root());
3699 Node *post = new AddINode(opaq, neg_stride);
3700 register_new_node(post, c);
3701 _igvn.rehash_node_delayed(use);
3702 for (uint j = 1; j < use->req(); j++) {
3703 if (use->in(j) == phi)
3704 use->set_req(j, post);
3705 }
3706 // Since DU info changed, rerun loop
3707 progress = true;
3708 break;
3709 }
3710 }
3711
3712 }