1 /*
2 * Copyright (c) 2009, 2023, 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 "ci/ciSymbols.hpp"
27 #include "classfile/javaClasses.hpp"
28 #include "compiler/compileLog.hpp"
29 #include "opto/callnode.hpp"
30 #include "opto/graphKit.hpp"
31 #include "opto/idealKit.hpp"
32 #include "opto/rootnode.hpp"
33 #include "opto/runtime.hpp"
34 #include "opto/stringopts.hpp"
35 #include "runtime/atomic.hpp"
36 #include "runtime/stubRoutines.hpp"
37
38 #define __ kit.
39
40 class StringConcat : public ResourceObj {
41 private:
42 PhaseStringOpts* _stringopts;
43 AllocateNode* _begin; // The allocation the begins the pattern
44 CallStaticJavaNode* _end; // The final call of the pattern. Will either be
45 // SB.toString or String.<init>(SB.toString)
46 bool _multiple; // indicates this is a fusion of two or more
47 // separate StringBuilders
48
49 Node* _arguments; // The list of arguments to be concatenated
50 GrowableArray<int> _mode; // into a String along with a mode flag
51 // indicating how to treat the value.
52 Node_List _constructors; // List of constructors (many in case of stacked concat)
53 Node_List _control; // List of control nodes that will be deleted
54 Node_List _uncommon_traps; // Uncommon traps that needs to be rewritten
55 // to restart at the initial JVMState.
56
57 public:
58 // Mode for converting arguments to Strings
59 enum {
60 StringMode,
61 IntMode,
62 CharMode,
63 StringNullCheckMode,
64 NegativeIntCheckMode
65 };
66
67 StringConcat(PhaseStringOpts* stringopts, CallStaticJavaNode* end):
68 _stringopts(stringopts),
69 _begin(nullptr),
70 _end(end),
71 _multiple(false) {
72 _arguments = new Node(1);
73 _arguments->del_req(0);
74 }
75
76 bool validate_mem_flow();
77 bool validate_control_flow();
78
79 StringConcat* merge(StringConcat* other, Node* arg);
80
81 void set_allocation(AllocateNode* alloc) {
82 _begin = alloc;
83 }
84
85 void append(Node* value, int mode) {
86 _arguments->add_req(value);
87 _mode.append(mode);
88 }
89 void push(Node* value, int mode) {
90 _arguments->ins_req(0, value);
91 _mode.insert_before(0, mode);
92 }
93
94 void push_string(Node* value) {
95 push(value, StringMode);
96 }
97
98 void push_string_null_check(Node* value) {
99 push(value, StringNullCheckMode);
100 }
101
102 void push_negative_int_check(Node* value) {
103 push(value, NegativeIntCheckMode);
104 }
105
106 void push_int(Node* value) {
107 push(value, IntMode);
108 }
109
110 void push_char(Node* value) {
111 push(value, CharMode);
112 }
113
114 static bool is_SB_toString(Node* call) {
115 if (call->is_CallStaticJava()) {
116 CallStaticJavaNode* csj = call->as_CallStaticJava();
117 ciMethod* m = csj->method();
118 if (m != nullptr &&
119 (m->intrinsic_id() == vmIntrinsics::_StringBuilder_toString ||
120 m->intrinsic_id() == vmIntrinsics::_StringBuffer_toString)) {
121 return true;
122 }
123 }
124 return false;
125 }
126
127 static Node* skip_string_null_check(Node* value) {
128 // Look for a diamond shaped Null check of toString() result
129 // (could be code from String.valueOf()):
130 // (Proj == nullptr) ? "null":"CastPP(Proj)#Notnull
131 if (value->is_Phi()) {
132 int true_path = value->as_Phi()->is_diamond_phi();
133 if (true_path != 0) {
134 // phi->region->if_proj->ifnode->bool
135 BoolNode* b = value->in(0)->in(1)->in(0)->in(1)->as_Bool();
136 Node* cmp = b->in(1);
137 Node* v1 = cmp->in(1);
138 Node* v2 = cmp->in(2);
139 // Null check of the return of toString which can simply be skipped.
140 if (b->_test._test == BoolTest::ne &&
141 v2->bottom_type() == TypePtr::NULL_PTR &&
142 value->in(true_path)->Opcode() == Op_CastPP &&
143 value->in(true_path)->in(1) == v1 &&
144 v1->is_Proj() && is_SB_toString(v1->in(0))) {
145 return v1;
146 }
147 }
148 }
149 return value;
150 }
151
152 Node* argument(int i) {
153 return _arguments->in(i);
154 }
155 Node* argument_uncast(int i) {
156 Node* arg = argument(i);
157 int amode = mode(i);
158 if (amode == StringConcat::StringMode ||
159 amode == StringConcat::StringNullCheckMode) {
160 arg = skip_string_null_check(arg);
161 }
162 return arg;
163 }
164 void set_argument(int i, Node* value) {
165 _arguments->set_req(i, value);
166 }
167 int num_arguments() {
168 return _mode.length();
169 }
170 int mode(int i) {
171 return _mode.at(i);
172 }
173 void add_control(Node* ctrl) {
174 assert(!_control.contains(ctrl), "only push once");
175 _control.push(ctrl);
176 }
177 void add_constructor(Node* init) {
178 assert(!_constructors.contains(init), "only push once");
179 _constructors.push(init);
180 }
181 CallStaticJavaNode* end() { return _end; }
182 AllocateNode* begin() { return _begin; }
183
184 void eliminate_unneeded_control();
185 void eliminate_initialize(InitializeNode* init);
186 void eliminate_call(CallNode* call);
187
188 void maybe_log_transform() {
189 CompileLog* log = _stringopts->C->log();
190 if (log != nullptr) {
191 log->head("replace_string_concat arguments='%d' multiple='%d'", num_arguments(), _multiple);
192 JVMState* p = _begin->jvms();
193 while (p != nullptr) {
194 log->elem("jvms bci='%d' method='%d'", p->bci(), log->identify(p->method()));
195 p = p->caller();
196 }
197 log->tail("replace_string_concat");
198 }
199 }
200
201 void convert_uncommon_traps(GraphKit& kit, const JVMState* jvms) {
202 for (uint u = 0; u < _uncommon_traps.size(); u++) {
203 Node* uct = _uncommon_traps.at(u);
204
205 // Build a new call using the jvms state of the allocate
206 address call_addr = OptoRuntime::uncommon_trap_blob()->entry_point();
207 const TypeFunc* call_type = OptoRuntime::uncommon_trap_Type();
208 const TypePtr* no_memory_effects = nullptr;
209 Compile* C = _stringopts->C;
210 CallStaticJavaNode* call = new CallStaticJavaNode(call_type, call_addr, "uncommon_trap",
211 no_memory_effects);
212 for (int e = 0; e < TypeFunc::Parms; e++) {
213 call->init_req(e, uct->in(e));
214 }
215 // Set the trap request to record intrinsic failure if this trap
216 // is taken too many times. Ideally we would handle then traps by
217 // doing the original bookkeeping in the MDO so that if it caused
218 // the code to be thrown out we could still recompile and use the
219 // optimization. Failing the uncommon traps doesn't really mean
220 // that the optimization is a bad idea but there's no other way to
221 // do the MDO updates currently.
222 int trap_request = Deoptimization::make_trap_request(Deoptimization::Reason_intrinsic,
223 Deoptimization::Action_make_not_entrant);
224 call->init_req(TypeFunc::Parms, __ intcon(trap_request));
225 kit.add_safepoint_edges(call);
226
227 _stringopts->gvn()->transform(call);
228 C->gvn_replace_by(uct, call);
229 uct->disconnect_inputs(C);
230 }
231 }
232
233 void cleanup() {
234 // disconnect the hook node
235 _arguments->disconnect_inputs(_stringopts->C);
236 }
237 };
238
239
240 void StringConcat::eliminate_unneeded_control() {
241 for (uint i = 0; i < _control.size(); i++) {
242 Node* n = _control.at(i);
243 if (n->is_Allocate()) {
244 eliminate_initialize(n->as_Allocate()->initialization());
245 }
246 if (n->is_Call()) {
247 if (n != _end) {
248 eliminate_call(n->as_Call());
249 }
250 } else if (n->is_IfTrue()) {
251 Compile* C = _stringopts->C;
252 C->gvn_replace_by(n, n->in(0)->in(0));
253 // get rid of the other projection
254 C->gvn_replace_by(n->in(0)->as_If()->proj_out(false), C->top());
255 } else if (n->is_Region()) {
256 Node* iff = n->in(1)->in(0);
257 assert(n->req() == 3 && n->in(2)->in(0) == iff, "not a diamond");
258 assert(iff->is_If(), "no if for the diamond");
259 Node* bol = iff->in(1);
260 assert(bol->is_Bool(), "unexpected if shape");
261 Node* cmp = bol->in(1);
262 assert(cmp->is_Cmp(), "unexpected if shape");
263 if (cmp->in(1)->is_top() || cmp->in(2)->is_top()) {
264 // This region should lose its Phis. They are removed either in PhaseRemoveUseless (for data phis) or in IGVN
265 // (for memory phis). During IGVN, there is a chance that the If folds to top before the Region is processed
266 // which then causes a reachable part of the graph to become dead. To prevent this, set the boolean input of
267 // the If to a constant to nicely let the diamond Region/If fold away.
268 Compile* C = _stringopts->C;
269 C->gvn_replace_by(iff->in(1), _stringopts->gvn()->intcon(0));
270 }
271 }
272 }
273 }
274
275
276 StringConcat* StringConcat::merge(StringConcat* other, Node* arg) {
277 StringConcat* result = new StringConcat(_stringopts, _end);
278 for (uint x = 0; x < _control.size(); x++) {
279 Node* n = _control.at(x);
280 if (n->is_Call()) {
281 result->_control.push(n);
282 }
283 }
284 for (uint x = 0; x < other->_control.size(); x++) {
285 Node* n = other->_control.at(x);
286 if (n->is_Call()) {
287 result->_control.push(n);
288 }
289 }
290 assert(result->_control.contains(other->_end), "what?");
291 assert(result->_control.contains(_begin), "what?");
292 for (int x = 0; x < num_arguments(); x++) {
293 Node* argx = argument_uncast(x);
294 if (argx == arg) {
295 // replace the toString result with the all the arguments that
296 // made up the other StringConcat
297 for (int y = 0; y < other->num_arguments(); y++) {
298 result->append(other->argument(y), other->mode(y));
299 }
300 } else {
301 result->append(argx, mode(x));
302 }
303 }
304 result->set_allocation(other->_begin);
305 for (uint i = 0; i < _constructors.size(); i++) {
306 result->add_constructor(_constructors.at(i));
307 }
308 for (uint i = 0; i < other->_constructors.size(); i++) {
309 result->add_constructor(other->_constructors.at(i));
310 }
311 result->_multiple = true;
312 return result;
313 }
314
315
316 void StringConcat::eliminate_call(CallNode* call) {
317 Compile* C = _stringopts->C;
318 CallProjections* projs = call->extract_projections(false);
319 if (projs->fallthrough_catchproj != nullptr) {
320 C->gvn_replace_by(projs->fallthrough_catchproj, call->in(TypeFunc::Control));
321 }
322 if (projs->fallthrough_memproj != nullptr) {
323 C->gvn_replace_by(projs->fallthrough_memproj, call->in(TypeFunc::Memory));
324 }
325 if (projs->catchall_memproj != nullptr) {
326 C->gvn_replace_by(projs->catchall_memproj, C->top());
327 }
328 if (projs->fallthrough_ioproj != nullptr) {
329 C->gvn_replace_by(projs->fallthrough_ioproj, call->in(TypeFunc::I_O));
330 }
331 if (projs->catchall_ioproj != nullptr) {
332 C->gvn_replace_by(projs->catchall_ioproj, C->top());
333 }
334 if (projs->catchall_catchproj != nullptr) {
335 // EA can't cope with the partially collapsed graph this
336 // creates so put it on the worklist to be collapsed later.
337 for (SimpleDUIterator i(projs->catchall_catchproj); i.has_next(); i.next()) {
338 Node *use = i.get();
339 int opc = use->Opcode();
340 if (opc == Op_CreateEx || opc == Op_Region) {
341 _stringopts->record_dead_node(use);
342 }
343 }
344 C->gvn_replace_by(projs->catchall_catchproj, C->top());
345 }
346 if (projs->resproj[0] != nullptr) {
347 assert(projs->nb_resproj == 1, "unexpected number of results");
348 C->gvn_replace_by(projs->resproj[0], C->top());
349 }
350 C->gvn_replace_by(call, C->top());
351 }
352
353 void StringConcat::eliminate_initialize(InitializeNode* init) {
354 Compile* C = _stringopts->C;
355
356 // Eliminate Initialize node.
357 assert(init->outcnt() <= 2, "only a control and memory projection expected");
358 assert(init->req() <= InitializeNode::RawStores, "no pending inits");
359 Node *ctrl_proj = init->proj_out_or_null(TypeFunc::Control);
360 if (ctrl_proj != nullptr) {
361 C->gvn_replace_by(ctrl_proj, init->in(TypeFunc::Control));
362 }
363 Node *mem_proj = init->proj_out_or_null(TypeFunc::Memory);
364 if (mem_proj != nullptr) {
365 Node *mem = init->in(TypeFunc::Memory);
366 C->gvn_replace_by(mem_proj, mem);
367 }
368 C->gvn_replace_by(init, C->top());
369 init->disconnect_inputs(C);
370 }
371
372 Node_List PhaseStringOpts::collect_toString_calls() {
373 Node_List string_calls;
374 Node_List worklist;
375
376 _visited.clear();
377
378 // Prime the worklist
379 for (uint i = 1; i < C->root()->len(); i++) {
380 Node* n = C->root()->in(i);
381 if (n != nullptr && !_visited.test_set(n->_idx)) {
382 worklist.push(n);
383 }
384 }
385
386 uint encountered = 0;
387 while (worklist.size() > 0) {
388 Node* ctrl = worklist.pop();
389 if (StringConcat::is_SB_toString(ctrl)) {
390 CallStaticJavaNode* csj = ctrl->as_CallStaticJava();
391 string_calls.push(csj);
392 encountered++;
393 }
394 if (ctrl->in(0) != nullptr && !_visited.test_set(ctrl->in(0)->_idx)) {
395 worklist.push(ctrl->in(0));
396 }
397 if (ctrl->is_Region()) {
398 for (uint i = 1; i < ctrl->len(); i++) {
399 if (ctrl->in(i) != nullptr && !_visited.test_set(ctrl->in(i)->_idx)) {
400 worklist.push(ctrl->in(i));
401 }
402 }
403 }
404 }
405 #ifndef PRODUCT
406 Atomic::add(&_stropts_total, encountered);
407 #endif
408 return string_calls;
409 }
410
411 // Recognize a fluent-chain of StringBuilder/Buffer. They are either explicit usages
412 // of them or the legacy bytecodes of string concatenation prior to JEP-280. eg.
413 //
414 // String result = new StringBuilder()
415 // .append("foo")
416 // .append("bar")
417 // .append(123)
418 // .toString(); // "foobar123"
419 //
420 // PS: Only a certain subset of constructor and append methods are acceptable.
421 // The criterion is that the length of argument is easy to work out in this phrase.
422 // It will drop complex cases such as Object.
423 //
424 // Since it walks along the receivers of fluent-chain, it will give up if the codeshape is
425 // not "fluent" enough. eg.
426 // StringBuilder sb = new StringBuilder();
427 // sb.append("foo");
428 // sb.toString();
429 //
430 // The receiver of toString method is the result of Allocation Node(CheckCastPP).
431 // The append method is overlooked. It will fail at validate_control_flow() test.
432 //
433 StringConcat* PhaseStringOpts::build_candidate(CallStaticJavaNode* call) {
434 ciMethod* m = call->method();
435 ciSymbol* string_sig;
436 ciSymbol* int_sig;
437 ciSymbol* char_sig;
438 if (m->holder() == C->env()->StringBuilder_klass()) {
439 string_sig = ciSymbols::String_StringBuilder_signature();
440 int_sig = ciSymbols::int_StringBuilder_signature();
441 char_sig = ciSymbols::char_StringBuilder_signature();
442 } else if (m->holder() == C->env()->StringBuffer_klass()) {
443 string_sig = ciSymbols::String_StringBuffer_signature();
444 int_sig = ciSymbols::int_StringBuffer_signature();
445 char_sig = ciSymbols::char_StringBuffer_signature();
446 } else {
447 return nullptr;
448 }
449 #ifndef PRODUCT
450 if (PrintOptimizeStringConcat) {
451 tty->print("considering toString call in ");
452 call->jvms()->dump_spec(tty); tty->cr();
453 }
454 #endif
455
456 StringConcat* sc = new StringConcat(this, call);
457 AllocateNode* alloc = nullptr;
458
459 // possible opportunity for StringBuilder fusion
460 CallStaticJavaNode* cnode = call;
461 while (cnode) {
462 Node* recv = cnode->in(TypeFunc::Parms)->uncast();
463 if (recv->is_Proj()) {
464 recv = recv->in(0);
465 }
466 cnode = recv->isa_CallStaticJava();
467 if (cnode == nullptr) {
468 alloc = recv->isa_Allocate();
469 if (alloc == nullptr) {
470 break;
471 }
472 // Find the constructor call
473 Node* result = alloc->result_cast();
474 if (result == nullptr || !result->is_CheckCastPP() || alloc->in(TypeFunc::Memory)->is_top()) {
475 // strange looking allocation
476 #ifndef PRODUCT
477 if (PrintOptimizeStringConcat) {
478 tty->print("giving up because allocation looks strange ");
479 alloc->jvms()->dump_spec(tty); tty->cr();
480 }
481 #endif
482 break;
483 }
484 Node* constructor = nullptr;
485 for (SimpleDUIterator i(result); i.has_next(); i.next()) {
486 CallStaticJavaNode *use = i.get()->isa_CallStaticJava();
487 if (use != nullptr &&
488 use->method() != nullptr &&
489 !use->method()->is_static() &&
490 use->method()->name() == ciSymbols::object_initializer_name() &&
491 use->method()->holder() == m->holder()) {
492 // Matched the constructor.
493 ciSymbol* sig = use->method()->signature()->as_symbol();
494 if (sig == ciSymbols::void_method_signature() ||
495 sig == ciSymbols::int_void_signature() ||
496 sig == ciSymbols::string_void_signature()) {
497 if (sig == ciSymbols::string_void_signature()) {
498 // StringBuilder(String) so pick this up as the first argument
499 assert(use->in(TypeFunc::Parms + 1) != nullptr, "what?");
500 const Type* type = _gvn->type(use->in(TypeFunc::Parms + 1));
501 if (type == TypePtr::NULL_PTR) {
502 // StringBuilder(null) throws exception.
503 #ifndef PRODUCT
504 if (PrintOptimizeStringConcat) {
505 tty->print("giving up because StringBuilder(null) throws exception");
506 alloc->jvms()->dump_spec(tty);
507 tty->cr();
508 }
509 #endif
510 return nullptr;
511 }
512 // StringBuilder(str) argument needs null check.
513 sc->push_string_null_check(use->in(TypeFunc::Parms + 1));
514 } else if (sig == ciSymbols::int_void_signature()) {
515 // StringBuilder(int) case.
516 Node* parm = use->in(TypeFunc::Parms + 1);
517 assert(parm != nullptr, "must exist");
518 const TypeInt* type = _gvn->type(parm)->is_int();
519 if (type->_hi < 0) {
520 // Initial capacity argument is always negative in which case StringBuilder(int) throws
521 // a NegativeArraySizeException. Bail out from string opts.
522 #ifndef PRODUCT
523 if (PrintOptimizeStringConcat) {
524 tty->print("giving up because a negative argument is passed to StringBuilder(int) which "
525 "throws a NegativeArraySizeException");
526 alloc->jvms()->dump_spec(tty);
527 tty->cr();
528 }
529 #endif
530 return nullptr;
531 } else if (type->_lo < 0) {
532 // Argument could be negative: We need a runtime check to throw NegativeArraySizeException in that case.
533 sc->push_negative_int_check(parm);
534 }
535 }
536 // The int variant takes an initial size for the backing
537 // array so just treat it like the void version.
538 constructor = use;
539 } else {
540 #ifndef PRODUCT
541 if (PrintOptimizeStringConcat) {
542 tty->print("unexpected constructor signature: %s", sig->as_utf8());
543 }
544 #endif
545 }
546 break;
547 }
548 }
549 if (constructor == nullptr) {
550 // couldn't find constructor
551 #ifndef PRODUCT
552 if (PrintOptimizeStringConcat) {
553 tty->print("giving up because couldn't find constructor ");
554 alloc->jvms()->dump_spec(tty); tty->cr();
555 }
556 #endif
557 break;
558 }
559
560 // Walked all the way back and found the constructor call so see
561 // if this call converted into a direct string concatenation.
562 sc->add_control(call);
563 sc->add_control(constructor);
564 sc->add_control(alloc);
565 sc->set_allocation(alloc);
566 sc->add_constructor(constructor);
567 if (sc->validate_control_flow() && sc->validate_mem_flow()) {
568 return sc;
569 } else {
570 return nullptr;
571 }
572 } else if (cnode->method() == nullptr) {
573 break;
574 } else if (!cnode->method()->is_static() &&
575 cnode->method()->holder() == m->holder() &&
576 cnode->method()->name() == ciSymbols::append_name() &&
577 (cnode->method()->signature()->as_symbol() == string_sig ||
578 cnode->method()->signature()->as_symbol() == char_sig ||
579 cnode->method()->signature()->as_symbol() == int_sig)) {
580 sc->add_control(cnode);
581 Node* arg = cnode->in(TypeFunc::Parms + 1);
582 if (arg == nullptr || arg->is_top()) {
583 #ifndef PRODUCT
584 if (PrintOptimizeStringConcat) {
585 tty->print("giving up because the call is effectively dead");
586 cnode->jvms()->dump_spec(tty); tty->cr();
587 }
588 #endif
589 break;
590 }
591 if (cnode->method()->signature()->as_symbol() == int_sig) {
592 sc->push_int(arg);
593 } else if (cnode->method()->signature()->as_symbol() == char_sig) {
594 sc->push_char(arg);
595 } else {
596 if (arg->is_Proj() && arg->in(0)->is_CallStaticJava()) {
597 CallStaticJavaNode* csj = arg->in(0)->as_CallStaticJava();
598 if (csj->method() != nullptr &&
599 csj->method()->intrinsic_id() == vmIntrinsics::_Integer_toString &&
600 arg->outcnt() == 1) {
601 // _control is the list of StringBuilder calls nodes which
602 // will be replaced by new String code after this optimization.
603 // Integer::toString() call is not part of StringBuilder calls
604 // chain. It could be eliminated only if its result is used
605 // only by this SB calls chain.
606 // Another limitation: it should be used only once because
607 // it is unknown that it is used only by this SB calls chain
608 // until all related SB calls nodes are collected.
609 assert(arg->unique_out() == cnode, "sanity");
610 sc->add_control(csj);
611 sc->push_int(csj->in(TypeFunc::Parms));
612 continue;
613 }
614 }
615 sc->push_string(arg);
616 }
617 continue;
618 } else {
619 // some unhandled signature
620 #ifndef PRODUCT
621 if (PrintOptimizeStringConcat) {
622 tty->print("giving up because encountered unexpected signature ");
623 cnode->tf()->dump(); tty->cr();
624 cnode->in(TypeFunc::Parms + 1)->dump();
625 }
626 #endif
627 break;
628 }
629 }
630 return nullptr;
631 }
632
633
634 PhaseStringOpts::PhaseStringOpts(PhaseGVN* gvn):
635 Phase(StringOpts),
636 _gvn(gvn) {
637
638 assert(OptimizeStringConcat, "shouldn't be here");
639
640 // Collect the types needed to talk about the various slices of memory
641 byte_adr_idx = C->get_alias_index(TypeAryPtr::BYTES);
642
643 // For each locally allocated StringBuffer see if the usages can be
644 // collapsed into a single String construction.
645
646 // Run through the list of allocation looking for SB.toString to see
647 // if it's possible to fuse the usage of the SB into a single String
648 // construction.
649 GrowableArray<StringConcat*> concats;
650 Node_List toStrings = collect_toString_calls();
651 while (toStrings.size() > 0) {
652 StringConcat* sc = build_candidate(toStrings.pop()->as_CallStaticJava());
653 if (sc != nullptr) {
654 concats.push(sc);
655 }
656 }
657
658 // try to coalesce separate concats
659 restart:
660 for (int c = 0; c < concats.length(); c++) {
661 StringConcat* sc = concats.at(c);
662 for (int i = 0; i < sc->num_arguments(); i++) {
663 Node* arg = sc->argument_uncast(i);
664 if (arg->is_Proj() && StringConcat::is_SB_toString(arg->in(0))) {
665 CallStaticJavaNode* csj = arg->in(0)->as_CallStaticJava();
666 for (int o = 0; o < concats.length(); o++) {
667 if (c == o) continue;
668 StringConcat* other = concats.at(o);
669 if (other->end() == csj) {
670 #ifndef PRODUCT
671 if (PrintOptimizeStringConcat) {
672 tty->print_cr("considering stacked concats");
673 }
674 #endif
675
676 StringConcat* merged = sc->merge(other, arg);
677 if (merged->validate_control_flow() && merged->validate_mem_flow()) {
678 #ifndef PRODUCT
679 Atomic::inc(&_stropts_merged);
680 if (PrintOptimizeStringConcat) {
681 tty->print_cr("stacking would succeed");
682 }
683 #endif
684 if (c < o) {
685 concats.remove_at(o);
686 concats.at_put(c, merged);
687 } else {
688 concats.remove_at(c);
689 concats.at_put(o, merged);
690 }
691 goto restart;
692 } else {
693 #ifndef PRODUCT
694 if (PrintOptimizeStringConcat) {
695 tty->print_cr("stacking would fail");
696 }
697 #endif
698 }
699 }
700 }
701 }
702 }
703 }
704
705
706 for (int c = 0; c < concats.length(); c++) {
707 StringConcat* sc = concats.at(c);
708 replace_string_concat(sc);
709 }
710
711 remove_dead_nodes();
712 }
713
714 void PhaseStringOpts::record_dead_node(Node* dead) {
715 dead_worklist.push(dead);
716 }
717
718 void PhaseStringOpts::remove_dead_nodes() {
719 // Delete any dead nodes to make things clean enough that escape
720 // analysis doesn't get unhappy.
721 while (dead_worklist.size() > 0) {
722 Node* use = dead_worklist.pop();
723 int opc = use->Opcode();
724 switch (opc) {
725 case Op_Region: {
726 uint i = 1;
727 for (i = 1; i < use->req(); i++) {
728 if (use->in(i) != C->top()) {
729 break;
730 }
731 }
732 if (i >= use->req()) {
733 for (SimpleDUIterator i(use); i.has_next(); i.next()) {
734 Node* m = i.get();
735 if (m->is_Phi()) {
736 dead_worklist.push(m);
737 }
738 }
739 C->gvn_replace_by(use, C->top());
740 }
741 break;
742 }
743 case Op_AddP:
744 case Op_CreateEx: {
745 // Recursively clean up references to CreateEx so EA doesn't
746 // get unhappy about the partially collapsed graph.
747 for (SimpleDUIterator i(use); i.has_next(); i.next()) {
748 Node* m = i.get();
749 if (m->is_AddP()) {
750 dead_worklist.push(m);
751 }
752 }
753 C->gvn_replace_by(use, C->top());
754 break;
755 }
756 case Op_Phi:
757 if (use->in(0) == C->top()) {
758 C->gvn_replace_by(use, C->top());
759 }
760 break;
761 }
762 }
763 }
764
765
766 bool StringConcat::validate_mem_flow() {
767 Compile* C = _stringopts->C;
768
769 for (uint i = 0; i < _control.size(); i++) {
770 #ifndef PRODUCT
771 Node_List path;
772 #endif
773 Node* curr = _control.at(i);
774 if (curr->is_Call() && curr != _begin) { // For all calls except the first allocation
775 // Now here's the main invariant in our case:
776 // For memory between the constructor, and appends, and toString we should only see bottom memory,
777 // produced by the previous call we know about.
778 if (!_constructors.contains(curr)) {
779 NOT_PRODUCT(path.push(curr);)
780 Node* mem = curr->in(TypeFunc::Memory);
781 assert(mem != nullptr, "calls should have memory edge");
782 assert(!mem->is_Phi(), "should be handled by control flow validation");
783 NOT_PRODUCT(path.push(mem);)
784 while (mem->is_MergeMem()) {
785 for (uint i = 1; i < mem->req(); i++) {
786 if (i != Compile::AliasIdxBot && mem->in(i) != C->top()) {
787 #ifndef PRODUCT
788 if (PrintOptimizeStringConcat) {
789 tty->print("fusion has incorrect memory flow (side effects) for ");
790 _begin->jvms()->dump_spec(tty); tty->cr();
791 path.dump();
792 }
793 #endif
794 return false;
795 }
796 }
797 // skip through a potential MergeMem chain, linked through Bot
798 mem = mem->in(Compile::AliasIdxBot);
799 NOT_PRODUCT(path.push(mem);)
800 }
801 // now let it fall through, and see if we have a projection
802 if (mem->is_Proj()) {
803 // Should point to a previous known call
804 Node *prev = mem->in(0);
805 NOT_PRODUCT(path.push(prev);)
806 if (!prev->is_Call() || !_control.contains(prev)) {
807 #ifndef PRODUCT
808 if (PrintOptimizeStringConcat) {
809 tty->print("fusion has incorrect memory flow (unknown call) for ");
810 _begin->jvms()->dump_spec(tty); tty->cr();
811 path.dump();
812 }
813 #endif
814 return false;
815 }
816 } else {
817 assert(mem->is_Store() || mem->is_LoadStore(), "unexpected node type: %s", mem->Name());
818 #ifndef PRODUCT
819 if (PrintOptimizeStringConcat) {
820 tty->print("fusion has incorrect memory flow (unexpected source) for ");
821 _begin->jvms()->dump_spec(tty); tty->cr();
822 path.dump();
823 }
824 #endif
825 return false;
826 }
827 } else {
828 // For memory that feeds into constructors it's more complicated.
829 // However the advantage is that any side effect that happens between the Allocate/Initialize and
830 // the constructor will have to be control-dependent on Initialize.
831 // So we actually don't have to do anything, since it's going to be caught by the control flow
832 // analysis.
833 #ifdef ASSERT
834 // Do a quick verification of the control pattern between the constructor and the initialize node
835 assert(curr->is_Call(), "constructor should be a call");
836 // Go up the control starting from the constructor call
837 Node* ctrl = curr->in(0);
838 IfNode* iff = nullptr;
839 RegionNode* copy = nullptr;
840
841 while (true) {
842 // skip known check patterns
843 if (ctrl->is_Region()) {
844 if (ctrl->as_Region()->is_copy()) {
845 copy = ctrl->as_Region();
846 ctrl = copy->is_copy();
847 } else { // a cast
848 assert(ctrl->req() == 3 &&
849 ctrl->in(1) != nullptr && ctrl->in(1)->is_Proj() &&
850 ctrl->in(2) != nullptr && ctrl->in(2)->is_Proj() &&
851 ctrl->in(1)->in(0) == ctrl->in(2)->in(0) &&
852 ctrl->in(1)->in(0) != nullptr && ctrl->in(1)->in(0)->is_If(),
853 "must be a simple diamond");
854 Node* true_proj = ctrl->in(1)->is_IfTrue() ? ctrl->in(1) : ctrl->in(2);
855 for (SimpleDUIterator i(true_proj); i.has_next(); i.next()) {
856 Node* use = i.get();
857 assert(use == ctrl || use->is_ConstraintCast(),
858 "unexpected user: %s", use->Name());
859 }
860
861 iff = ctrl->in(1)->in(0)->as_If();
862 ctrl = iff->in(0);
863 }
864 } else if (ctrl->is_IfTrue()) { // null checks, class checks
865 iff = ctrl->in(0)->as_If();
866 // Verify that the other arm is an uncommon trap
867 Node* otherproj = iff->proj_out(1 - ctrl->as_Proj()->_con);
868 CallStaticJavaNode* call = otherproj->unique_out()->isa_CallStaticJava();
869 assert(strcmp(call->_name, "uncommon_trap") == 0, "must be uncommon trap");
870 ctrl = iff->in(0);
871 } else {
872 break;
873 }
874 }
875
876 assert(ctrl->is_Proj(), "must be a projection");
877 assert(ctrl->in(0)->is_Initialize(), "should be initialize");
878 for (SimpleDUIterator i(ctrl); i.has_next(); i.next()) {
879 Node* use = i.get();
880 assert(use == copy || use == iff || use == curr || use->is_CheckCastPP() || use->is_Load(),
881 "unexpected user: %s", use->Name());
882 }
883 #endif // ASSERT
884 }
885 }
886 }
887
888 #ifndef PRODUCT
889 if (PrintOptimizeStringConcat) {
890 tty->print("fusion has correct memory flow for ");
891 _begin->jvms()->dump_spec(tty); tty->cr();
892 tty->cr();
893 }
894 #endif
895 return true;
896 }
897
898 bool StringConcat::validate_control_flow() {
899 // We found all the calls and arguments now lets see if it's
900 // safe to transform the graph as we would expect.
901
902 // Check to see if this resulted in too many uncommon traps previously
903 if (Compile::current()->too_many_traps(_begin->jvms()->method(), _begin->jvms()->bci(),
904 Deoptimization::Reason_intrinsic)) {
905 return false;
906 }
907
908 // Walk backwards over the control flow from toString to the
909 // allocation and make sure all the control flow is ok. This
910 // means it's either going to be eliminated once the calls are
911 // removed or it can safely be transformed into an uncommon
912 // trap.
913
914 int null_check_count = 0;
915 Unique_Node_List ctrl_path;
916
917 assert(_control.contains(_begin), "missing");
918 assert(_control.contains(_end), "missing");
919
920 // Collect the nodes that we know about and will eliminate into ctrl_path
921 for (uint i = 0; i < _control.size(); i++) {
922 // Push the call and it's control projection
923 Node* n = _control.at(i);
924 if (n->is_Allocate()) {
925 AllocateNode* an = n->as_Allocate();
926 InitializeNode* init = an->initialization();
927 ctrl_path.push(init);
928 ctrl_path.push(init->as_Multi()->proj_out(0));
929 }
930 if (n->is_Call()) {
931 CallNode* cn = n->as_Call();
932 ctrl_path.push(cn);
933 ctrl_path.push(cn->proj_out(0));
934 ctrl_path.push(cn->proj_out(0)->unique_out());
935 Node* catchproj = cn->proj_out(0)->unique_out()->as_Catch()->proj_out_or_null(0);
936 if (catchproj != nullptr) {
937 ctrl_path.push(catchproj);
938 }
939 } else {
940 ShouldNotReachHere();
941 }
942 }
943
944 // Skip backwards through the control checking for unexpected control flow
945 Node* ptr = _end;
946 bool fail = false;
947 while (ptr != _begin) {
948 if (ptr->is_Call() && ctrl_path.member(ptr)) {
949 ptr = ptr->in(0);
950 } else if (ptr->is_CatchProj() && ctrl_path.member(ptr)) {
951 ptr = ptr->in(0)->in(0)->in(0);
952 assert(ctrl_path.member(ptr), "should be a known piece of control");
953 } else if (ptr->is_IfTrue()) {
954 IfNode* iff = ptr->in(0)->as_If();
955 BoolNode* b = iff->in(1)->isa_Bool();
956
957 if (b == nullptr) {
958 #ifndef PRODUCT
959 if (PrintOptimizeStringConcat) {
960 tty->print_cr("unexpected input to IfNode");
961 iff->in(1)->dump();
962 tty->cr();
963 }
964 #endif
965 fail = true;
966 break;
967 }
968
969 Node* cmp = b->in(1);
970 Node* v1 = cmp->in(1);
971 Node* v2 = cmp->in(2);
972 Node* otherproj = iff->proj_out(1 - ptr->as_Proj()->_con);
973
974 // Null check of the return of append which can simply be eliminated
975 if (b->_test._test == BoolTest::ne &&
976 v2->bottom_type() == TypePtr::NULL_PTR &&
977 v1->is_Proj() && ctrl_path.member(v1->in(0))) {
978 // null check of the return value of the append
979 null_check_count++;
980 if (otherproj->outcnt() == 1) {
981 CallStaticJavaNode* call = otherproj->unique_out()->isa_CallStaticJava();
982 if (call != nullptr && call->_name != nullptr && strcmp(call->_name, "uncommon_trap") == 0) {
983 ctrl_path.push(call);
984 }
985 }
986 _control.push(ptr);
987 ptr = ptr->in(0)->in(0);
988 continue;
989 }
990
991 // A test which leads to an uncommon trap which should be safe.
992 // Later this trap will be converted into a trap that restarts
993 // at the beginning.
994 if (otherproj->outcnt() == 1) {
995 CallStaticJavaNode* call = otherproj->unique_out()->isa_CallStaticJava();
996 if (call != nullptr && call->_name != nullptr && strcmp(call->_name, "uncommon_trap") == 0) {
997 // control flow leads to uct so should be ok
998 _uncommon_traps.push(call);
999 ctrl_path.push(call);
1000 ptr = ptr->in(0)->in(0);
1001 continue;
1002 }
1003 }
1004
1005 #ifndef PRODUCT
1006 // Some unexpected control flow we don't know how to handle.
1007 if (PrintOptimizeStringConcat) {
1008 tty->print_cr("failing with unknown test");
1009 b->dump();
1010 cmp->dump();
1011 v1->dump();
1012 v2->dump();
1013 tty->cr();
1014 }
1015 #endif
1016 fail = true;
1017 break;
1018 } else if (ptr->is_Proj() && ptr->in(0)->is_Initialize()) {
1019 // Check for side effect between Initialize and the constructor
1020 for (SimpleDUIterator iter(ptr); iter.has_next(); iter.next()) {
1021 Node* use = iter.get();
1022 if (!use->is_CFG() && !use->is_CheckCastPP() && !use->is_Load()) {
1023 #ifndef PRODUCT
1024 if (PrintOptimizeStringConcat) {
1025 tty->print_cr("unexpected control use of Initialize");
1026 ptr->in(0)->dump(); // Initialize node
1027 use->dump(1);
1028 }
1029 #endif
1030 fail = true;
1031 break;
1032 }
1033 }
1034 ptr = ptr->in(0)->in(0);
1035 } else if (ptr->is_Region()) {
1036 Node* copy = ptr->as_Region()->is_copy();
1037 if (copy != nullptr) {
1038 ptr = copy;
1039 continue;
1040 }
1041 if (ptr->req() == 3 &&
1042 ptr->in(1) != nullptr && ptr->in(1)->is_Proj() &&
1043 ptr->in(2) != nullptr && ptr->in(2)->is_Proj() &&
1044 ptr->in(1)->in(0) == ptr->in(2)->in(0) &&
1045 ptr->in(1)->in(0) != nullptr && ptr->in(1)->in(0)->is_If()) {
1046 // Simple diamond.
1047 // XXX should check for possibly merging stores. simple data merges are ok.
1048 // The IGVN will make this simple diamond go away when it
1049 // transforms the Region. Make sure it sees it.
1050 Compile::current()->record_for_igvn(ptr);
1051 _control.push(ptr);
1052 ptr = ptr->in(1)->in(0)->in(0);
1053 continue;
1054 }
1055 #ifndef PRODUCT
1056 if (PrintOptimizeStringConcat) {
1057 tty->print_cr("fusion would fail for region");
1058 _begin->dump();
1059 ptr->dump(2);
1060 }
1061 #endif
1062 fail = true;
1063 break;
1064 } else {
1065 // other unknown control
1066 if (!fail) {
1067 #ifndef PRODUCT
1068 if (PrintOptimizeStringConcat) {
1069 tty->print_cr("fusion would fail for");
1070 _begin->dump();
1071 }
1072 #endif
1073 fail = true;
1074 }
1075 #ifndef PRODUCT
1076 if (PrintOptimizeStringConcat) {
1077 ptr->dump();
1078 }
1079 #endif
1080 ptr = ptr->in(0);
1081 }
1082 }
1083 #ifndef PRODUCT
1084 if (PrintOptimizeStringConcat && fail) {
1085 tty->cr();
1086 }
1087 #endif
1088 if (fail) return !fail;
1089
1090 // Validate that all these results produced are contained within
1091 // this cluster of objects. First collect all the results produced
1092 // by calls in the region.
1093 _stringopts->_visited.clear();
1094 Node_List worklist;
1095 Node* final_result = _end->proj_out_or_null(TypeFunc::Parms);
1096 for (uint i = 0; i < _control.size(); i++) {
1097 CallNode* cnode = _control.at(i)->isa_Call();
1098 if (cnode != nullptr) {
1099 _stringopts->_visited.test_set(cnode->_idx);
1100 }
1101 Node* result = cnode != nullptr ? cnode->proj_out_or_null(TypeFunc::Parms) : nullptr;
1102 if (result != nullptr && result != final_result) {
1103 worklist.push(result);
1104 }
1105 }
1106
1107 Node* last_result = nullptr;
1108 while (worklist.size() > 0) {
1109 Node* result = worklist.pop();
1110 if (_stringopts->_visited.test_set(result->_idx))
1111 continue;
1112 for (SimpleDUIterator i(result); i.has_next(); i.next()) {
1113 Node *use = i.get();
1114 if (ctrl_path.member(use)) {
1115 // already checked this
1116 continue;
1117 }
1118 int opc = use->Opcode();
1119 if (opc == Op_CmpP || opc == Op_Node) {
1120 ctrl_path.push(use);
1121 continue;
1122 }
1123 if (opc == Op_CastPP || opc == Op_CheckCastPP) {
1124 for (SimpleDUIterator j(use); j.has_next(); j.next()) {
1125 worklist.push(j.get());
1126 }
1127 worklist.push(use->in(1));
1128 ctrl_path.push(use);
1129 continue;
1130 }
1131 #ifndef PRODUCT
1132 if (PrintOptimizeStringConcat) {
1133 if (result != last_result) {
1134 last_result = result;
1135 tty->print_cr("extra uses for result:");
1136 last_result->dump();
1137 }
1138 use->dump();
1139 }
1140 #endif
1141 fail = true;
1142 break;
1143 }
1144 }
1145
1146 #ifndef PRODUCT
1147 if (PrintOptimizeStringConcat && !fail) {
1148 ttyLocker ttyl;
1149 tty->cr();
1150 tty->print("fusion has correct control flow (%d %d) for ", null_check_count, _uncommon_traps.size());
1151 _begin->jvms()->dump_spec(tty); tty->cr();
1152 for (int i = 0; i < num_arguments(); i++) {
1153 argument(i)->dump();
1154 }
1155 _control.dump();
1156 tty->cr();
1157 }
1158 #endif
1159
1160 return !fail;
1161 }
1162
1163 // Mirror of Integer.stringSize() method, return the count of digits in integer,
1164 Node* PhaseStringOpts::int_stringSize(GraphKit& kit, Node* arg) {
1165 if (arg->is_Con()) {
1166 // Constant integer. Compute constant length
1167 jint arg_val = arg->get_int();
1168 jint d = 1;
1169 if (arg_val >= 0) {
1170 d = 0;
1171 arg_val = -arg_val;
1172 }
1173 jint p = -10;
1174 for (int i = 1; i < 10; i++) {
1175 if (arg_val > p) {
1176 return __ intcon(i + d);
1177 }
1178 p = java_multiply(10, p);
1179 }
1180 return __ intcon(10 + d);
1181 }
1182
1183 // int d = 1;
1184 // if (x >= 0) {
1185 // d = 0;
1186 // x = -x;
1187 // }
1188 RegionNode* sign_merge = new RegionNode(3);
1189 kit.gvn().set_type(sign_merge, Type::CONTROL);
1190 Node* digit_cnt = new PhiNode(sign_merge, TypeInt::INT);
1191 kit.gvn().set_type(digit_cnt, TypeInt::INT);
1192 Node* val = new PhiNode(sign_merge, TypeInt::INT);
1193 kit.gvn().set_type(val, TypeInt::INT);
1194
1195 IfNode* iff = kit.create_and_map_if(kit.control(),
1196 __ Bool(__ CmpI(arg, __ intcon(0)), BoolTest::ge),
1197 PROB_FAIR, COUNT_UNKNOWN);
1198 sign_merge->init_req(1, __ IfTrue(iff));
1199 sign_merge->init_req(2, __ IfFalse(iff));
1200 digit_cnt->init_req(1, __ intcon(0));
1201 digit_cnt->init_req(2, __ intcon(1));
1202 val->init_req(1, __ SubI(__ intcon(0), arg));
1203 val->init_req(2, arg);
1204 kit.set_control(sign_merge);
1205
1206 // int p = -10;
1207 // for (int i = 1; i < 10; i++) {
1208 // if (x > p)
1209 // return i + d;
1210 // p = 10 * p;
1211 // }
1212 RegionNode* final_merge = new RegionNode(3);
1213 kit.gvn().set_type(final_merge, Type::CONTROL);
1214 Node* final_size = new PhiNode(final_merge, TypeInt::INT);
1215 kit.gvn().set_type(final_size, TypeInt::INT);
1216
1217 kit.add_parse_predicates();
1218 C->set_has_loops(true);
1219
1220 RegionNode* loop = new RegionNode(3);
1221 kit.gvn().set_type(loop, Type::CONTROL);
1222 Node* index = new PhiNode(loop, TypeInt::INT);
1223 kit.gvn().set_type(index, TypeInt::INT);
1224 Node* temp = new PhiNode(loop, TypeInt::INT);
1225 kit.gvn().set_type(temp, TypeInt::INT);
1226
1227 loop->init_req(1, kit.control());
1228 index->init_req(1, __ intcon(1));
1229 temp->init_req(1, __ intcon(-10));
1230 kit.set_control(loop);
1231
1232 Node* limit = __ CmpI(index, __ intcon(10));
1233 Node* limitb = __ Bool(limit, BoolTest::lt);
1234 IfNode* iff2 = kit.create_and_map_if(kit.control(), limitb, PROB_MIN, COUNT_UNKNOWN);
1235 Node* limit_less = __ IfTrue(iff2);
1236 kit.set_control(limit_less);
1237
1238 Node* cmp = __ CmpI(val, temp);
1239 Node* cmpb = __ Bool(cmp, BoolTest::gt);
1240 IfNode* iff3 = kit.create_and_map_if(kit.control(), cmpb, PROB_MIN, COUNT_UNKNOWN);
1241 Node* cmp_le = __ IfFalse(iff3);
1242 kit.set_control(cmp_le);
1243
1244 loop->init_req(2, kit.control());
1245 index->init_req(2, __ AddI(index, __ intcon(1)));
1246 temp->init_req(2, __ MulI(temp, __ intcon(10)));
1247
1248 final_merge->init_req(1, __ IfFalse(iff2));
1249 final_merge->init_req(2, __ IfTrue(iff3));
1250 final_size->init_req(1, __ AddI(digit_cnt, __ intcon(10)));
1251 final_size->init_req(2, __ AddI(digit_cnt, index));
1252 kit.set_control(final_merge);
1253
1254 C->record_for_igvn(sign_merge);
1255 C->record_for_igvn(digit_cnt);
1256 C->record_for_igvn(val);
1257 C->record_for_igvn(final_merge);
1258 C->record_for_igvn(final_size);
1259 C->record_for_igvn(loop);
1260 C->record_for_igvn(index);
1261 C->record_for_igvn(temp);
1262 return final_size;
1263 }
1264
1265 // Simplified version of Integer.getChars
1266 void PhaseStringOpts::getChars(GraphKit& kit, Node* arg, Node* dst_array, BasicType bt, Node* end, Node* final_merge, Node* final_mem, int merge_index) {
1267 // if (i < 0) {
1268 // sign = '-';
1269 // i = -i;
1270 // }
1271 IfNode* iff = kit.create_and_map_if(kit.control(), __ Bool(__ CmpI(arg, __ intcon(0)), BoolTest::lt),
1272 PROB_FAIR, COUNT_UNKNOWN);
1273
1274 RegionNode* merge = new RegionNode(3);
1275 kit.gvn().set_type(merge, Type::CONTROL);
1276 Node* i = new PhiNode(merge, TypeInt::INT);
1277 kit.gvn().set_type(i, TypeInt::INT);
1278 Node* sign = new PhiNode(merge, TypeInt::INT);
1279 kit.gvn().set_type(sign, TypeInt::INT);
1280
1281 merge->init_req(1, __ IfTrue(iff));
1282 i->init_req(1, __ SubI(__ intcon(0), arg));
1283 sign->init_req(1, __ intcon('-'));
1284 merge->init_req(2, __ IfFalse(iff));
1285 i->init_req(2, arg);
1286 sign->init_req(2, __ intcon(0));
1287
1288 kit.set_control(merge);
1289
1290 C->record_for_igvn(merge);
1291 C->record_for_igvn(i);
1292 C->record_for_igvn(sign);
1293
1294 // for (;;) {
1295 // q = i / 10;
1296 // r = i - ((q << 3) + (q << 1)); // r = i-(q*10) ...
1297 // buf [--charPos] = digits [r];
1298 // i = q;
1299 // if (i == 0) break;
1300 // }
1301
1302 // Add Parse Predicates first.
1303 kit.add_parse_predicates();
1304
1305 C->set_has_loops(true);
1306 RegionNode* head = new RegionNode(3);
1307 head->init_req(1, kit.control());
1308
1309 kit.gvn().set_type(head, Type::CONTROL);
1310 Node* i_phi = new PhiNode(head, TypeInt::INT);
1311 i_phi->init_req(1, i);
1312 kit.gvn().set_type(i_phi, TypeInt::INT);
1313 Node* charPos = new PhiNode(head, TypeInt::INT);
1314 charPos->init_req(1, end);
1315 kit.gvn().set_type(charPos, TypeInt::INT);
1316 Node* mem = PhiNode::make(head, kit.memory(byte_adr_idx), Type::MEMORY, TypeAryPtr::BYTES);
1317 kit.gvn().set_type(mem, Type::MEMORY);
1318
1319 kit.set_control(head);
1320 kit.set_memory(mem, byte_adr_idx);
1321
1322 Node* q = __ DivI(kit.null(), i_phi, __ intcon(10));
1323 Node* r = __ SubI(i_phi, __ AddI(__ LShiftI(q, __ intcon(3)),
1324 __ LShiftI(q, __ intcon(1))));
1325 Node* index = __ SubI(charPos, __ intcon((bt == T_BYTE) ? 1 : 2));
1326 Node* ch = __ AddI(r, __ intcon('0'));
1327 Node* st = __ store_to_memory(kit.control(), kit.array_element_address(dst_array, index, T_BYTE),
1328 ch, bt, byte_adr_idx, MemNode::unordered, false /* require_atomic_access */,
1329 false /* unaligned */, (bt != T_BYTE) /* mismatched */);
1330
1331 iff = kit.create_and_map_if(head, __ Bool(__ CmpI(q, __ intcon(0)), BoolTest::ne),
1332 PROB_FAIR, COUNT_UNKNOWN);
1333 Node* ne = __ IfTrue(iff);
1334 Node* eq = __ IfFalse(iff);
1335
1336 head->init_req(2, ne);
1337 mem->init_req(2, st);
1338
1339 i_phi->init_req(2, q);
1340 charPos->init_req(2, index);
1341 charPos = index;
1342
1343 kit.set_control(eq);
1344 kit.set_memory(st, byte_adr_idx);
1345
1346 C->record_for_igvn(head);
1347 C->record_for_igvn(mem);
1348 C->record_for_igvn(i_phi);
1349 C->record_for_igvn(charPos);
1350
1351 // if (sign != 0) {
1352 // buf [--charPos] = sign;
1353 // }
1354 iff = kit.create_and_map_if(kit.control(), __ Bool(__ CmpI(sign, __ intcon(0)), BoolTest::ne),
1355 PROB_FAIR, COUNT_UNKNOWN);
1356
1357 final_merge->init_req(merge_index + 2, __ IfFalse(iff));
1358 final_mem->init_req(merge_index + 2, kit.memory(byte_adr_idx));
1359
1360 kit.set_control(__ IfTrue(iff));
1361 if (kit.stopped()) {
1362 final_merge->init_req(merge_index + 1, C->top());
1363 final_mem->init_req(merge_index + 1, C->top());
1364 } else {
1365 Node* index = __ SubI(charPos, __ intcon((bt == T_BYTE) ? 1 : 2));
1366 st = __ store_to_memory(kit.control(), kit.array_element_address(dst_array, index, T_BYTE),
1367 sign, bt, byte_adr_idx, MemNode::unordered, false /* require_atomic_access */,
1368 false /* unaligned */, (bt != T_BYTE) /* mismatched */);
1369
1370 final_merge->init_req(merge_index + 1, kit.control());
1371 final_mem->init_req(merge_index + 1, st);
1372 }
1373 }
1374
1375 // Copy the characters representing arg into dst_array starting at start
1376 Node* PhaseStringOpts::int_getChars(GraphKit& kit, Node* arg, Node* dst_array, Node* dst_coder, Node* start, Node* size) {
1377 bool dcon = dst_coder->is_Con();
1378 bool dbyte = dcon ? (dst_coder->get_int() == java_lang_String::CODER_LATIN1) : false;
1379 Node* end = __ AddI(start, __ LShiftI(size, dst_coder));
1380
1381 // The final_merge node has 4 entries in case the encoding is known:
1382 // (0) Control, (1) result w/ sign, (2) result w/o sign, (3) result for Integer.min_value
1383 // or 6 entries in case the encoding is not known:
1384 // (0) Control, (1) Latin1 w/ sign, (2) Latin1 w/o sign, (3) min_value, (4) UTF16 w/ sign, (5) UTF16 w/o sign
1385 RegionNode* final_merge = new RegionNode(dcon ? 4 : 6);
1386 kit.gvn().set_type(final_merge, Type::CONTROL);
1387
1388 Node* final_mem = PhiNode::make(final_merge, kit.memory(byte_adr_idx), Type::MEMORY, TypeAryPtr::BYTES);
1389 kit.gvn().set_type(final_mem, Type::MEMORY);
1390
1391 // need to handle arg == Integer.MIN_VALUE specially because negating doesn't make it positive
1392 IfNode* iff = kit.create_and_map_if(kit.control(), __ Bool(__ CmpI(arg, __ intcon(0x80000000)), BoolTest::ne),
1393 PROB_FAIR, COUNT_UNKNOWN);
1394
1395 Node* old_mem = kit.memory(byte_adr_idx);
1396
1397 kit.set_control(__ IfFalse(iff));
1398 if (kit.stopped()) {
1399 // Statically not equal to MIN_VALUE so this path is dead
1400 final_merge->init_req(3, kit.control());
1401 } else {
1402 copy_string(kit, __ makecon(TypeInstPtr::make(C->env()->the_min_jint_string())),
1403 dst_array, dst_coder, start);
1404 final_merge->init_req(3, kit.control());
1405 final_mem->init_req(3, kit.memory(byte_adr_idx));
1406 }
1407
1408 kit.set_control(__ IfTrue(iff));
1409 kit.set_memory(old_mem, byte_adr_idx);
1410
1411 if (!dcon) {
1412 // Check encoding of destination
1413 iff = kit.create_and_map_if(kit.control(), __ Bool(__ CmpI(dst_coder, __ intcon(0)), BoolTest::eq),
1414 PROB_FAIR, COUNT_UNKNOWN);
1415 old_mem = kit.memory(byte_adr_idx);
1416 }
1417 if (!dcon || dbyte) {
1418 // Destination is Latin1,
1419 if (!dcon) {
1420 kit.set_control(__ IfTrue(iff));
1421 }
1422 getChars(kit, arg, dst_array, T_BYTE, end, final_merge, final_mem);
1423 }
1424 if (!dcon || !dbyte) {
1425 // Destination is UTF16
1426 int merge_index = 0;
1427 if (!dcon) {
1428 kit.set_control(__ IfFalse(iff));
1429 kit.set_memory(old_mem, byte_adr_idx);
1430 merge_index = 3; // Account for Latin1 case
1431 }
1432 getChars(kit, arg, dst_array, T_CHAR, end, final_merge, final_mem, merge_index);
1433 }
1434
1435 // Final merge point for Latin1 and UTF16 case
1436 kit.set_control(final_merge);
1437 kit.set_memory(final_mem, byte_adr_idx);
1438
1439 C->record_for_igvn(final_merge);
1440 C->record_for_igvn(final_mem);
1441 return end;
1442 }
1443
1444 // Copy 'count' bytes/chars from src_array to dst_array starting at index start
1445 void PhaseStringOpts::arraycopy(GraphKit& kit, IdealKit& ideal, Node* src_array, Node* dst_array, BasicType elembt, Node* start, Node* count) {
1446 assert(elembt == T_BYTE || elembt == T_CHAR, "Invalid type for arraycopy");
1447
1448 if (elembt == T_CHAR) {
1449 // Get number of chars
1450 count = __ RShiftI(count, __ intcon(1));
1451 }
1452
1453 Node* extra = nullptr;
1454 #ifdef _LP64
1455 count = __ ConvI2L(count);
1456 extra = C->top();
1457 #endif
1458
1459 Node* src_ptr = __ array_element_address(src_array, __ intcon(0), T_BYTE);
1460 Node* dst_ptr = __ array_element_address(dst_array, start, T_BYTE);
1461 // Check if destination address is aligned to HeapWordSize
1462 const TypeInt* tdst = __ gvn().type(start)->is_int();
1463 bool aligned = tdst->is_con() && ((tdst->get_con() * type2aelembytes(T_BYTE)) % HeapWordSize == 0);
1464 // Figure out which arraycopy runtime method to call (disjoint, uninitialized).
1465 const char* copyfunc_name = "arraycopy";
1466 address copyfunc_addr = StubRoutines::select_arraycopy_function(elembt, aligned, true, copyfunc_name, true);
1467 ideal.make_leaf_call_no_fp(OptoRuntime::fast_arraycopy_Type(), copyfunc_addr, copyfunc_name,
1468 TypeAryPtr::BYTES, src_ptr, dst_ptr, count, extra);
1469 }
1470
1471 #undef __
1472 #define __ ideal.
1473
1474 // Copy contents of a Latin1 encoded string from src_array to dst_array
1475 void PhaseStringOpts::copy_latin1_string(GraphKit& kit, IdealKit& ideal, Node* src_array, IdealVariable& count,
1476 Node* dst_array, Node* dst_coder, Node* start) {
1477 bool dcon = dst_coder->is_Con();
1478 bool dbyte = dcon ? (dst_coder->get_int() == java_lang_String::CODER_LATIN1) : false;
1479
1480 if (!dcon) {
1481 __ if_then(dst_coder, BoolTest::eq, __ ConI(java_lang_String::CODER_LATIN1));
1482 }
1483 if (!dcon || dbyte) {
1484 // Destination is Latin1. Simply emit a byte arraycopy.
1485 arraycopy(kit, ideal, src_array, dst_array, T_BYTE, start, __ value(count));
1486 }
1487 if (!dcon) {
1488 __ else_();
1489 }
1490 if (!dcon || !dbyte) {
1491 // Destination is UTF16. Inflate src_array into dst_array.
1492 kit.sync_kit(ideal);
1493 if (Matcher::match_rule_supported(Op_StrInflatedCopy)) {
1494 // Use fast intrinsic
1495 Node* src = kit.array_element_address(src_array, kit.intcon(0), T_BYTE);
1496 Node* dst = kit.array_element_address(dst_array, start, T_BYTE);
1497 kit.inflate_string(src, dst, TypeAryPtr::BYTES, __ value(count));
1498 } else {
1499 // No intrinsic available, use slow method
1500 kit.inflate_string_slow(src_array, dst_array, start, __ value(count));
1501 }
1502 ideal.sync_kit(&kit);
1503 // Multiply count by two since we now need two bytes per char
1504 __ set(count, __ LShiftI(__ value(count), __ ConI(1)));
1505 }
1506 if (!dcon) {
1507 __ end_if();
1508 }
1509 }
1510
1511 // Read two bytes from index and index+1 and convert them to a char
1512 static jchar readChar(ciTypeArray* array, int index) {
1513 int shift_high, shift_low;
1514 #ifdef VM_LITTLE_ENDIAN
1515 shift_high = 0;
1516 shift_low = 8;
1517 #else
1518 shift_high = 8;
1519 shift_low = 0;
1520 #endif
1521
1522 jchar b1 = ((jchar) array->byte_at(index)) & 0xff;
1523 jchar b2 = ((jchar) array->byte_at(index+1)) & 0xff;
1524 return (b1 << shift_high) | (b2 << shift_low);
1525 }
1526
1527 // Copy contents of constant src_array to dst_array by emitting individual stores
1528 void PhaseStringOpts::copy_constant_string(GraphKit& kit, IdealKit& ideal, ciTypeArray* src_array, IdealVariable& count,
1529 bool src_is_byte, Node* dst_array, Node* dst_coder, Node* start) {
1530 bool dcon = dst_coder->is_Con();
1531 bool dbyte = dcon ? (dst_coder->get_int() == java_lang_String::CODER_LATIN1) : false;
1532 int length = src_array->length();
1533
1534 if (!dcon) {
1535 __ if_then(dst_coder, BoolTest::eq, __ ConI(java_lang_String::CODER_LATIN1));
1536 }
1537 if (!dcon || dbyte) {
1538 // Destination is Latin1. Copy each byte of src_array into dst_array.
1539 Node* index = start;
1540 for (int i = 0; i < length; i++) {
1541 Node* adr = kit.array_element_address(dst_array, index, T_BYTE);
1542 Node* val = __ ConI(src_array->byte_at(i));
1543 __ store(__ ctrl(), adr, val, T_BYTE, byte_adr_idx, MemNode::unordered);
1544 index = __ AddI(index, __ ConI(1));
1545 }
1546 }
1547 if (!dcon) {
1548 __ else_();
1549 }
1550 if (!dcon || !dbyte) {
1551 // Destination is UTF16. Copy each char of src_array into dst_array.
1552 Node* index = start;
1553 for (int i = 0; i < length; i++) {
1554 Node* adr = kit.array_element_address(dst_array, index, T_BYTE);
1555 jchar val;
1556 if (src_is_byte) {
1557 val = src_array->byte_at(i) & 0xff;
1558 } else {
1559 val = readChar(src_array, i++);
1560 }
1561 __ store(__ ctrl(), adr, __ ConI(val), T_CHAR, byte_adr_idx, MemNode::unordered, false /* require_atomic_access */,
1562 true /* mismatched */);
1563 index = __ AddI(index, __ ConI(2));
1564 }
1565 if (src_is_byte) {
1566 // Multiply count by two since we now need two bytes per char
1567 __ set(count, __ ConI(2 * length));
1568 }
1569 }
1570 if (!dcon) {
1571 __ end_if();
1572 }
1573 }
1574
1575 // Compress copy contents of the byte/char String str into dst_array starting at index start.
1576 Node* PhaseStringOpts::copy_string(GraphKit& kit, Node* str, Node* dst_array, Node* dst_coder, Node* start) {
1577 Node* src_array = kit.load_String_value(str, true);
1578
1579 IdealKit ideal(&kit, true, true);
1580 IdealVariable count(ideal); __ declarations_done();
1581
1582 if (str->is_Con()) {
1583 // Constant source string
1584 ciTypeArray* src_array_type = get_constant_value(kit, str);
1585
1586 // Check encoding of constant string
1587 bool src_is_byte = (get_constant_coder(kit, str) == java_lang_String::CODER_LATIN1);
1588
1589 // For small constant strings just emit individual stores.
1590 // A length of 6 seems like a good space/speed tradeof.
1591 __ set(count, __ ConI(src_array_type->length()));
1592 int src_len = src_array_type->length() / (src_is_byte ? 1 : 2);
1593 if (src_len < unroll_string_copy_length) {
1594 // Small constant string
1595 copy_constant_string(kit, ideal, src_array_type, count, src_is_byte, dst_array, dst_coder, start);
1596 } else if (src_is_byte) {
1597 // Source is Latin1
1598 copy_latin1_string(kit, ideal, src_array, count, dst_array, dst_coder, start);
1599 } else {
1600 // Source is UTF16 (destination too). Simply emit a char arraycopy.
1601 arraycopy(kit, ideal, src_array, dst_array, T_CHAR, start, __ value(count));
1602 }
1603 } else {
1604 Node* size = kit.load_array_length(src_array);
1605 __ set(count, size);
1606 // Non-constant source string
1607 if (CompactStrings) {
1608 // Emit runtime check for coder
1609 Node* coder = kit.load_String_coder(str, true);
1610 __ if_then(coder, BoolTest::eq, __ ConI(java_lang_String::CODER_LATIN1)); {
1611 // Source is Latin1
1612 copy_latin1_string(kit, ideal, src_array, count, dst_array, dst_coder, start);
1613 } __ else_();
1614 }
1615 // Source is UTF16 (destination too). Simply emit a char arraycopy.
1616 arraycopy(kit, ideal, src_array, dst_array, T_CHAR, start, __ value(count));
1617
1618 if (CompactStrings) {
1619 __ end_if();
1620 }
1621 }
1622
1623 // Finally sync IdealKit and GraphKit.
1624 kit.sync_kit(ideal);
1625 return __ AddI(start, __ value(count));
1626 }
1627
1628 // Compress copy the char into dst_array at index start.
1629 Node* PhaseStringOpts::copy_char(GraphKit& kit, Node* val, Node* dst_array, Node* dst_coder, Node* start) {
1630 bool dcon = (dst_coder != nullptr) && dst_coder->is_Con();
1631 bool dbyte = dcon ? (dst_coder->get_int() == java_lang_String::CODER_LATIN1) : false;
1632
1633 IdealKit ideal(&kit, true, true);
1634 IdealVariable end(ideal); __ declarations_done();
1635 Node* adr = kit.array_element_address(dst_array, start, T_BYTE);
1636 if (!dcon){
1637 __ if_then(dst_coder, BoolTest::eq, __ ConI(java_lang_String::CODER_LATIN1));
1638 }
1639 if (!dcon || dbyte) {
1640 // Destination is Latin1. Store a byte.
1641 __ store(__ ctrl(), adr, val, T_BYTE, byte_adr_idx, MemNode::unordered);
1642 __ set(end, __ AddI(start, __ ConI(1)));
1643 }
1644 if (!dcon) {
1645 __ else_();
1646 }
1647 if (!dcon || !dbyte) {
1648 // Destination is UTF16. Store a char.
1649 __ store(__ ctrl(), adr, val, T_CHAR, byte_adr_idx, MemNode::unordered, false /* require_atomic_access */,
1650 true /* mismatched */);
1651 __ set(end, __ AddI(start, __ ConI(2)));
1652 }
1653 if (!dcon) {
1654 __ end_if();
1655 }
1656 // Finally sync IdealKit and GraphKit.
1657 kit.sync_kit(ideal);
1658 return __ value(end);
1659 }
1660
1661 #undef __
1662 #define __ kit.
1663
1664 // Allocate a byte array of specified length.
1665 Node* PhaseStringOpts::allocate_byte_array(GraphKit& kit, IdealKit* ideal, Node* length) {
1666 if (ideal != nullptr) {
1667 // Sync IdealKit and graphKit.
1668 kit.sync_kit(*ideal);
1669 }
1670 Node* byte_array = nullptr;
1671 {
1672 PreserveReexecuteState preexecs(&kit);
1673 // The original jvms is for an allocation of either a String or
1674 // StringBuffer so no stack adjustment is necessary for proper
1675 // reexecution. If we deoptimize in the slow path the bytecode
1676 // will be reexecuted and the char[] allocation will be thrown away.
1677 kit.jvms()->set_should_reexecute(true);
1678 byte_array = kit.new_array(__ makecon(TypeKlassPtr::make(ciTypeArrayKlass::make(T_BYTE))),
1679 length, 1);
1680 }
1681
1682 // Mark the allocation so that zeroing is skipped since the code
1683 // below will overwrite the entire array
1684 AllocateArrayNode* byte_alloc = AllocateArrayNode::Ideal_array_allocation(byte_array);
1685 byte_alloc->maybe_set_complete(_gvn);
1686
1687 if (ideal != nullptr) {
1688 // Sync IdealKit and graphKit.
1689 ideal->sync_kit(&kit);
1690 }
1691 return byte_array;
1692 }
1693
1694 jbyte PhaseStringOpts::get_constant_coder(GraphKit& kit, Node* str) {
1695 assert(str->is_Con(), "String must be constant");
1696 const TypeOopPtr* str_type = kit.gvn().type(str)->isa_oopptr();
1697 ciInstance* str_instance = str_type->const_oop()->as_instance();
1698 jbyte coder = str_instance->field_value_by_offset(java_lang_String::coder_offset()).as_byte();
1699 assert(CompactStrings || (coder == java_lang_String::CODER_UTF16), "Strings must be UTF16 encoded");
1700 return coder;
1701 }
1702
1703 int PhaseStringOpts::get_constant_length(GraphKit& kit, Node* str) {
1704 assert(str->is_Con(), "String must be constant");
1705 return get_constant_value(kit, str)->length();
1706 }
1707
1708 ciTypeArray* PhaseStringOpts::get_constant_value(GraphKit& kit, Node* str) {
1709 assert(str->is_Con(), "String must be constant");
1710 const TypeOopPtr* str_type = kit.gvn().type(str)->isa_oopptr();
1711 ciInstance* str_instance = str_type->const_oop()->as_instance();
1712 ciObject* src_array = str_instance->field_value_by_offset(java_lang_String::value_offset()).as_object();
1713 return src_array->as_type_array();
1714 }
1715
1716 void PhaseStringOpts::replace_string_concat(StringConcat* sc) {
1717 // Log a little info about the transformation
1718 sc->maybe_log_transform();
1719
1720 // pull the JVMState of the allocation into a SafePointNode to serve as
1721 // as a shim for the insertion of the new code.
1722 JVMState* jvms = sc->begin()->jvms()->clone_shallow(C);
1723 uint size = sc->begin()->req();
1724 SafePointNode* map = new SafePointNode(size, jvms);
1725
1726 // copy the control and memory state from the final call into our
1727 // new starting state. This allows any preceding tests to feed
1728 // into the new section of code.
1729 for (uint i1 = 0; i1 < TypeFunc::Parms; i1++) {
1730 map->init_req(i1, sc->end()->in(i1));
1731 }
1732 // blow away old allocation arguments
1733 for (uint i1 = TypeFunc::Parms; i1 < jvms->debug_start(); i1++) {
1734 map->init_req(i1, C->top());
1735 }
1736 // Copy the rest of the inputs for the JVMState
1737 for (uint i1 = jvms->debug_start(); i1 < sc->begin()->req(); i1++) {
1738 map->init_req(i1, sc->begin()->in(i1));
1739 }
1740 // Make sure the memory state is a MergeMem for parsing.
1741 if (!map->in(TypeFunc::Memory)->is_MergeMem()) {
1742 map->set_req(TypeFunc::Memory, MergeMemNode::make(map->in(TypeFunc::Memory)));
1743 }
1744
1745 jvms->set_map(map);
1746 map->ensure_stack(jvms, jvms->method()->max_stack());
1747
1748 // disconnect all the old StringBuilder calls from the graph
1749 sc->eliminate_unneeded_control();
1750
1751 // At this point all the old work has been completely removed from
1752 // the graph and the saved JVMState exists at the point where the
1753 // final toString call used to be.
1754 GraphKit kit(jvms);
1755
1756 // There may be uncommon traps which are still using the
1757 // intermediate states and these need to be rewritten to point at
1758 // the JVMState at the beginning of the transformation.
1759 sc->convert_uncommon_traps(kit, jvms);
1760
1761 // Now insert the logic to compute the size of the string followed
1762 // by all the logic to construct array and resulting string.
1763
1764 Node* null_string = __ makecon(TypeInstPtr::make(C->env()->the_null_string()));
1765
1766 // Create a region for the overflow checks to merge into.
1767 int args = MAX2(sc->num_arguments(), 1);
1768 RegionNode* overflow = new RegionNode(args);
1769 kit.gvn().set_type(overflow, Type::CONTROL);
1770
1771 // Create a hook node to hold onto the individual sizes since they
1772 // are need for the copying phase.
1773 Node* string_sizes = new Node(args);
1774
1775 Node* coder = __ intcon(0);
1776 Node* length = __ intcon(0);
1777 // If at least one argument is UTF16 encoded, we can fix the encoding.
1778 bool coder_fixed = false;
1779
1780 if (!CompactStrings) {
1781 // Fix encoding of result string to UTF16
1782 coder_fixed = true;
1783 coder = __ intcon(java_lang_String::CODER_UTF16);
1784 }
1785
1786 for (int argi = 0; argi < sc->num_arguments(); argi++) {
1787 Node* arg = sc->argument(argi);
1788 switch (sc->mode(argi)) {
1789 case StringConcat::NegativeIntCheckMode: {
1790 // Initial capacity argument might be negative in which case StringBuilder(int) throws
1791 // a NegativeArraySizeException. Insert a runtime check with an uncommon trap.
1792 const TypeInt* type = kit.gvn().type(arg)->is_int();
1793 assert(type->_hi >= 0 && type->_lo < 0, "no runtime int check needed");
1794 Node* p = __ Bool(__ CmpI(arg, kit.intcon(0)), BoolTest::ge);
1795 IfNode* iff = kit.create_and_map_if(kit.control(), p, PROB_MIN, COUNT_UNKNOWN);
1796 {
1797 // Negative int -> uncommon trap.
1798 PreserveJVMState pjvms(&kit);
1799 kit.set_control(__ IfFalse(iff));
1800 kit.uncommon_trap(Deoptimization::Reason_intrinsic,
1801 Deoptimization::Action_maybe_recompile);
1802 }
1803 kit.set_control(__ IfTrue(iff));
1804 break;
1805 }
1806 case StringConcat::IntMode: {
1807 Node* string_size = int_stringSize(kit, arg);
1808
1809 // accumulate total
1810 length = __ AddI(length, string_size);
1811
1812 // Cache this value for the use by int_toString
1813 string_sizes->init_req(argi, string_size);
1814 break;
1815 }
1816 case StringConcat::StringNullCheckMode: {
1817 const Type* type = kit.gvn().type(arg);
1818 assert(type != TypePtr::NULL_PTR, "missing check");
1819 if (!type->higher_equal(TypeInstPtr::NOTNULL)) {
1820 // Null check with uncommon trap since
1821 // StringBuilder(null) throws exception.
1822 // Use special uncommon trap instead of
1823 // calling normal do_null_check().
1824 Node* p = __ Bool(__ CmpP(arg, kit.null()), BoolTest::ne);
1825 IfNode* iff = kit.create_and_map_if(kit.control(), p, PROB_MIN, COUNT_UNKNOWN);
1826 overflow->add_req(__ IfFalse(iff));
1827 Node* notnull = __ IfTrue(iff);
1828 kit.set_control(notnull); // set control for the cast_not_null
1829 arg = kit.cast_not_null(arg, false);
1830 sc->set_argument(argi, arg);
1831 }
1832 assert(kit.gvn().type(arg)->higher_equal(TypeInstPtr::NOTNULL), "sanity");
1833 // Fallthrough to add string length.
1834 }
1835 case StringConcat::StringMode: {
1836 const Type* type = kit.gvn().type(arg);
1837 Node* count = nullptr;
1838 Node* arg_coder = nullptr;
1839 if (type == TypePtr::NULL_PTR) {
1840 // replace the argument with the null checked version
1841 arg = null_string;
1842 sc->set_argument(argi, arg);
1843 count = kit.load_String_length(arg, true);
1844 arg_coder = kit.load_String_coder(arg, true);
1845 } else if (!type->higher_equal(TypeInstPtr::NOTNULL)) {
1846 // s = s != null ? s : "null";
1847 // length = length + (s.count - s.offset);
1848 RegionNode *r = new RegionNode(3);
1849 kit.gvn().set_type(r, Type::CONTROL);
1850 Node *phi = new PhiNode(r, type);
1851 kit.gvn().set_type(phi, phi->bottom_type());
1852 Node* p = __ Bool(__ CmpP(arg, kit.null()), BoolTest::ne);
1853 IfNode* iff = kit.create_and_map_if(kit.control(), p, PROB_MIN, COUNT_UNKNOWN);
1854 Node* notnull = __ IfTrue(iff);
1855 Node* isnull = __ IfFalse(iff);
1856 kit.set_control(notnull); // set control for the cast_not_null
1857 r->init_req(1, notnull);
1858 phi->init_req(1, kit.cast_not_null(arg, false));
1859 r->init_req(2, isnull);
1860 phi->init_req(2, null_string);
1861 kit.set_control(r);
1862 C->record_for_igvn(r);
1863 C->record_for_igvn(phi);
1864 // replace the argument with the null checked version
1865 arg = phi;
1866 sc->set_argument(argi, arg);
1867 count = kit.load_String_length(arg, true);
1868 arg_coder = kit.load_String_coder(arg, true);
1869 } else {
1870 // A corresponding nullcheck will be connected during IGVN MemNode::Ideal_common_DU_postCCP
1871 // kit.control might be a different test, that can be hoisted above the actual nullcheck
1872 // in case, that the control input is not null, Ideal_common_DU_postCCP will not look for a nullcheck.
1873 count = kit.load_String_length(arg, false);
1874 arg_coder = kit.load_String_coder(arg, false);
1875 }
1876 if (arg->is_Con()) {
1877 // Constant string. Get constant coder and length.
1878 jbyte const_coder = get_constant_coder(kit, arg);
1879 int const_length = get_constant_length(kit, arg);
1880 if (const_coder == java_lang_String::CODER_LATIN1) {
1881 // Can be latin1 encoded
1882 arg_coder = __ intcon(const_coder);
1883 count = __ intcon(const_length);
1884 } else {
1885 // Found UTF16 encoded string. Fix result array encoding to UTF16.
1886 coder_fixed = true;
1887 coder = __ intcon(const_coder);
1888 count = __ intcon(const_length / 2);
1889 }
1890 }
1891
1892 if (!coder_fixed) {
1893 coder = __ OrI(coder, arg_coder);
1894 }
1895 length = __ AddI(length, count);
1896 string_sizes->init_req(argi, nullptr);
1897 break;
1898 }
1899 case StringConcat::CharMode: {
1900 // one character only
1901 const TypeInt* t = kit.gvn().type(arg)->is_int();
1902 if (!coder_fixed && t->is_con()) {
1903 // Constant char
1904 if (t->get_con() <= 255) {
1905 // Can be latin1 encoded
1906 coder = __ OrI(coder, __ intcon(java_lang_String::CODER_LATIN1));
1907 } else {
1908 // Must be UTF16 encoded. Fix result array encoding to UTF16.
1909 coder_fixed = true;
1910 coder = __ intcon(java_lang_String::CODER_UTF16);
1911 }
1912 } else if (!coder_fixed) {
1913 // Not constant
1914 #undef __
1915 #define __ ideal.
1916 IdealKit ideal(&kit, true, true);
1917 IdealVariable char_coder(ideal); __ declarations_done();
1918 // Check if character can be latin1 encoded
1919 __ if_then(arg, BoolTest::le, __ ConI(0xFF));
1920 __ set(char_coder, __ ConI(java_lang_String::CODER_LATIN1));
1921 __ else_();
1922 __ set(char_coder, __ ConI(java_lang_String::CODER_UTF16));
1923 __ end_if();
1924 kit.sync_kit(ideal);
1925 coder = __ OrI(coder, __ value(char_coder));
1926 #undef __
1927 #define __ kit.
1928 }
1929 length = __ AddI(length, __ intcon(1));
1930 break;
1931 }
1932 default:
1933 ShouldNotReachHere();
1934 }
1935 if (argi > 0) {
1936 // Check that the sum hasn't overflowed
1937 IfNode* iff = kit.create_and_map_if(kit.control(),
1938 __ Bool(__ CmpI(length, __ intcon(0)), BoolTest::lt),
1939 PROB_MIN, COUNT_UNKNOWN);
1940 kit.set_control(__ IfFalse(iff));
1941 overflow->set_req(argi, __ IfTrue(iff));
1942 }
1943 }
1944
1945 {
1946 // Hook
1947 PreserveJVMState pjvms(&kit);
1948 kit.set_control(overflow);
1949 C->record_for_igvn(overflow);
1950 kit.uncommon_trap(Deoptimization::Reason_intrinsic,
1951 Deoptimization::Action_make_not_entrant);
1952 }
1953
1954 Node* result;
1955 if (!kit.stopped()) {
1956 assert(CompactStrings || (coder->is_Con() && coder->get_int() == java_lang_String::CODER_UTF16),
1957 "Result string must be UTF16 encoded if CompactStrings is disabled");
1958
1959 Node* dst_array = nullptr;
1960 if (sc->num_arguments() == 1 &&
1961 (sc->mode(0) == StringConcat::StringMode ||
1962 sc->mode(0) == StringConcat::StringNullCheckMode)) {
1963 // Handle the case when there is only a single String argument.
1964 // In this case, we can just pull the value from the String itself.
1965 dst_array = kit.load_String_value(sc->argument(0), true);
1966 } else {
1967 // Allocate destination byte array according to coder
1968 dst_array = allocate_byte_array(kit, nullptr, __ LShiftI(length, coder));
1969
1970 // Now copy the string representations into the final byte[]
1971 Node* start = __ intcon(0);
1972 for (int argi = 0; argi < sc->num_arguments(); argi++) {
1973 Node* arg = sc->argument(argi);
1974 switch (sc->mode(argi)) {
1975 case StringConcat::NegativeIntCheckMode:
1976 break; // Nothing to do, was only needed to add a runtime check earlier.
1977 case StringConcat::IntMode: {
1978 start = int_getChars(kit, arg, dst_array, coder, start, string_sizes->in(argi));
1979 break;
1980 }
1981 case StringConcat::StringNullCheckMode:
1982 case StringConcat::StringMode: {
1983 start = copy_string(kit, arg, dst_array, coder, start);
1984 break;
1985 }
1986 case StringConcat::CharMode: {
1987 start = copy_char(kit, arg, dst_array, coder, start);
1988 break;
1989 }
1990 default:
1991 ShouldNotReachHere();
1992 }
1993 }
1994 }
1995
1996 {
1997 PreserveReexecuteState preexecs(&kit);
1998 // The original jvms is for an allocation of either a String or
1999 // StringBuffer so no stack adjustment is necessary for proper
2000 // reexecution.
2001 kit.jvms()->set_should_reexecute(true);
2002 result = kit.new_instance(__ makecon(TypeKlassPtr::make(C->env()->String_klass())));
2003 }
2004
2005 // Initialize the string
2006 kit.store_String_value(result, dst_array);
2007 kit.store_String_coder(result, coder);
2008
2009 // The value field is final. Emit a barrier here to ensure that the effect
2010 // of the initialization is committed to memory before any code publishes
2011 // a reference to the newly constructed object (see Parse::do_exits()).
2012 assert(AllocateNode::Ideal_allocation(result) != nullptr, "should be newly allocated");
2013 kit.insert_mem_bar(UseStoreStoreForCtor ? Op_MemBarStoreStore : Op_MemBarRelease, result);
2014 } else {
2015 result = C->top();
2016 }
2017 // hook up the outgoing control and result
2018 kit.replace_call(sc->end(), result);
2019
2020 // Unhook any hook nodes
2021 string_sizes->disconnect_inputs(C);
2022 sc->cleanup();
2023 #ifndef PRODUCT
2024 Atomic::inc(&_stropts_replaced);
2025 #endif
2026 }
2027
2028 #ifndef PRODUCT
2029 uint PhaseStringOpts::_stropts_replaced = 0;
2030 uint PhaseStringOpts::_stropts_merged = 0;
2031 uint PhaseStringOpts::_stropts_total = 0;
2032
2033 void PhaseStringOpts::print_statistics() {
2034 tty->print_cr("StringConcat: %4d/%4d/%4d(replaced/merged/total)", _stropts_replaced, _stropts_merged, _stropts_total);
2035 }
2036 #endif