1 /* 2 * Copyright (c) 1997, 2011, 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 "asm/assembler.hpp" 27 #include "classfile/systemDictionary.hpp" 28 #include "code/exceptionHandlerTable.hpp" 29 #include "code/nmethod.hpp" 30 #include "compiler/compileLog.hpp" 31 #include "compiler/oopMap.hpp" 32 #include "opto/addnode.hpp" 33 #include "opto/block.hpp" 34 #include "opto/c2compiler.hpp" 35 #include "opto/callGenerator.hpp" 36 #include "opto/callnode.hpp" 37 #include "opto/cfgnode.hpp" 38 #include "opto/chaitin.hpp" 39 #include "opto/compile.hpp" 40 #include "opto/connode.hpp" 41 #include "opto/divnode.hpp" 42 #include "opto/escape.hpp" 43 #include "opto/idealGraphPrinter.hpp" 44 #include "opto/loopnode.hpp" 45 #include "opto/machnode.hpp" 46 #include "opto/macro.hpp" 47 #include "opto/matcher.hpp" 48 #include "opto/memnode.hpp" 49 #include "opto/mulnode.hpp" 50 #include "opto/node.hpp" 51 #include "opto/opcodes.hpp" 52 #include "opto/output.hpp" 53 #include "opto/parse.hpp" 54 #include "opto/phaseX.hpp" 55 #include "opto/rootnode.hpp" 56 #include "opto/runtime.hpp" 57 #include "opto/stringopts.hpp" 58 #include "opto/type.hpp" 59 #include "opto/vectornode.hpp" 60 #include "runtime/arguments.hpp" 61 #include "runtime/signature.hpp" 62 #include "runtime/stubRoutines.hpp" 63 #include "runtime/timer.hpp" 64 #include "utilities/copy.hpp" 65 #ifdef TARGET_ARCH_MODEL_x86_32 66 # include "adfiles/ad_x86_32.hpp" 67 #endif 68 #ifdef TARGET_ARCH_MODEL_x86_64 69 # include "adfiles/ad_x86_64.hpp" 70 #endif 71 #ifdef TARGET_ARCH_MODEL_sparc 72 # include "adfiles/ad_sparc.hpp" 73 #endif 74 #ifdef TARGET_ARCH_MODEL_zero 75 # include "adfiles/ad_zero.hpp" 76 #endif 77 #ifdef TARGET_ARCH_MODEL_arm 78 # include "adfiles/ad_arm.hpp" 79 #endif 80 #ifdef TARGET_ARCH_MODEL_ppc 81 # include "adfiles/ad_ppc.hpp" 82 #endif 83 84 85 // -------------------- Compile::mach_constant_base_node ----------------------- 86 // Constant table base node singleton. 87 MachConstantBaseNode* Compile::mach_constant_base_node() { 88 if (_mach_constant_base_node == NULL) { 89 _mach_constant_base_node = new (C) MachConstantBaseNode(); 90 _mach_constant_base_node->add_req(C->root()); 91 } 92 return _mach_constant_base_node; 93 } 94 95 96 /// Support for intrinsics. 97 98 // Return the index at which m must be inserted (or already exists). 99 // The sort order is by the address of the ciMethod, with is_virtual as minor key. 100 int Compile::intrinsic_insertion_index(ciMethod* m, bool is_virtual) { 101 #ifdef ASSERT 102 for (int i = 1; i < _intrinsics->length(); i++) { 103 CallGenerator* cg1 = _intrinsics->at(i-1); 104 CallGenerator* cg2 = _intrinsics->at(i); 105 assert(cg1->method() != cg2->method() 106 ? cg1->method() < cg2->method() 107 : cg1->is_virtual() < cg2->is_virtual(), 108 "compiler intrinsics list must stay sorted"); 109 } 110 #endif 111 // Binary search sorted list, in decreasing intervals [lo, hi]. 112 int lo = 0, hi = _intrinsics->length()-1; 113 while (lo <= hi) { 114 int mid = (uint)(hi + lo) / 2; 115 ciMethod* mid_m = _intrinsics->at(mid)->method(); 116 if (m < mid_m) { 117 hi = mid-1; 118 } else if (m > mid_m) { 119 lo = mid+1; 120 } else { 121 // look at minor sort key 122 bool mid_virt = _intrinsics->at(mid)->is_virtual(); 123 if (is_virtual < mid_virt) { 124 hi = mid-1; 125 } else if (is_virtual > mid_virt) { 126 lo = mid+1; 127 } else { 128 return mid; // exact match 129 } 130 } 131 } 132 return lo; // inexact match 133 } 134 135 void Compile::register_intrinsic(CallGenerator* cg) { 136 if (_intrinsics == NULL) { 137 _intrinsics = new GrowableArray<CallGenerator*>(60); 138 } 139 // This code is stolen from ciObjectFactory::insert. 140 // Really, GrowableArray should have methods for 141 // insert_at, remove_at, and binary_search. 142 int len = _intrinsics->length(); 143 int index = intrinsic_insertion_index(cg->method(), cg->is_virtual()); 144 if (index == len) { 145 _intrinsics->append(cg); 146 } else { 147 #ifdef ASSERT 148 CallGenerator* oldcg = _intrinsics->at(index); 149 assert(oldcg->method() != cg->method() || oldcg->is_virtual() != cg->is_virtual(), "don't register twice"); 150 #endif 151 _intrinsics->append(_intrinsics->at(len-1)); 152 int pos; 153 for (pos = len-2; pos >= index; pos--) { 154 _intrinsics->at_put(pos+1,_intrinsics->at(pos)); 155 } 156 _intrinsics->at_put(index, cg); 157 } 158 assert(find_intrinsic(cg->method(), cg->is_virtual()) == cg, "registration worked"); 159 } 160 161 CallGenerator* Compile::find_intrinsic(ciMethod* m, bool is_virtual) { 162 assert(m->is_loaded(), "don't try this on unloaded methods"); 163 if (_intrinsics != NULL) { 164 int index = intrinsic_insertion_index(m, is_virtual); 165 if (index < _intrinsics->length() 166 && _intrinsics->at(index)->method() == m 167 && _intrinsics->at(index)->is_virtual() == is_virtual) { 168 return _intrinsics->at(index); 169 } 170 } 171 // Lazily create intrinsics for intrinsic IDs well-known in the runtime. 172 if (m->intrinsic_id() != vmIntrinsics::_none && 173 m->intrinsic_id() <= vmIntrinsics::LAST_COMPILER_INLINE) { 174 CallGenerator* cg = make_vm_intrinsic(m, is_virtual); 175 if (cg != NULL) { 176 // Save it for next time: 177 register_intrinsic(cg); 178 return cg; 179 } else { 180 gather_intrinsic_statistics(m->intrinsic_id(), is_virtual, _intrinsic_disabled); 181 } 182 } 183 return NULL; 184 } 185 186 // Compile:: register_library_intrinsics and make_vm_intrinsic are defined 187 // in library_call.cpp. 188 189 190 #ifndef PRODUCT 191 // statistics gathering... 192 193 juint Compile::_intrinsic_hist_count[vmIntrinsics::ID_LIMIT] = {0}; 194 jubyte Compile::_intrinsic_hist_flags[vmIntrinsics::ID_LIMIT] = {0}; 195 196 bool Compile::gather_intrinsic_statistics(vmIntrinsics::ID id, bool is_virtual, int flags) { 197 assert(id > vmIntrinsics::_none && id < vmIntrinsics::ID_LIMIT, "oob"); 198 int oflags = _intrinsic_hist_flags[id]; 199 assert(flags != 0, "what happened?"); 200 if (is_virtual) { 201 flags |= _intrinsic_virtual; 202 } 203 bool changed = (flags != oflags); 204 if ((flags & _intrinsic_worked) != 0) { 205 juint count = (_intrinsic_hist_count[id] += 1); 206 if (count == 1) { 207 changed = true; // first time 208 } 209 // increment the overall count also: 210 _intrinsic_hist_count[vmIntrinsics::_none] += 1; 211 } 212 if (changed) { 213 if (((oflags ^ flags) & _intrinsic_virtual) != 0) { 214 // Something changed about the intrinsic's virtuality. 215 if ((flags & _intrinsic_virtual) != 0) { 216 // This is the first use of this intrinsic as a virtual call. 217 if (oflags != 0) { 218 // We already saw it as a non-virtual, so note both cases. 219 flags |= _intrinsic_both; 220 } 221 } else if ((oflags & _intrinsic_both) == 0) { 222 // This is the first use of this intrinsic as a non-virtual 223 flags |= _intrinsic_both; 224 } 225 } 226 _intrinsic_hist_flags[id] = (jubyte) (oflags | flags); 227 } 228 // update the overall flags also: 229 _intrinsic_hist_flags[vmIntrinsics::_none] |= (jubyte) flags; 230 return changed; 231 } 232 233 static char* format_flags(int flags, char* buf) { 234 buf[0] = 0; 235 if ((flags & Compile::_intrinsic_worked) != 0) strcat(buf, ",worked"); 236 if ((flags & Compile::_intrinsic_failed) != 0) strcat(buf, ",failed"); 237 if ((flags & Compile::_intrinsic_disabled) != 0) strcat(buf, ",disabled"); 238 if ((flags & Compile::_intrinsic_virtual) != 0) strcat(buf, ",virtual"); 239 if ((flags & Compile::_intrinsic_both) != 0) strcat(buf, ",nonvirtual"); 240 if (buf[0] == 0) strcat(buf, ","); 241 assert(buf[0] == ',', "must be"); 242 return &buf[1]; 243 } 244 245 void Compile::print_intrinsic_statistics() { 246 char flagsbuf[100]; 247 ttyLocker ttyl; 248 if (xtty != NULL) xtty->head("statistics type='intrinsic'"); 249 tty->print_cr("Compiler intrinsic usage:"); 250 juint total = _intrinsic_hist_count[vmIntrinsics::_none]; 251 if (total == 0) total = 1; // avoid div0 in case of no successes 252 #define PRINT_STAT_LINE(name, c, f) \ 253 tty->print_cr(" %4d (%4.1f%%) %s (%s)", (int)(c), ((c) * 100.0) / total, name, f); 254 for (int index = 1 + (int)vmIntrinsics::_none; index < (int)vmIntrinsics::ID_LIMIT; index++) { 255 vmIntrinsics::ID id = (vmIntrinsics::ID) index; 256 int flags = _intrinsic_hist_flags[id]; 257 juint count = _intrinsic_hist_count[id]; 258 if ((flags | count) != 0) { 259 PRINT_STAT_LINE(vmIntrinsics::name_at(id), count, format_flags(flags, flagsbuf)); 260 } 261 } 262 PRINT_STAT_LINE("total", total, format_flags(_intrinsic_hist_flags[vmIntrinsics::_none], flagsbuf)); 263 if (xtty != NULL) xtty->tail("statistics"); 264 } 265 266 void Compile::print_statistics() { 267 { ttyLocker ttyl; 268 if (xtty != NULL) xtty->head("statistics type='opto'"); 269 Parse::print_statistics(); 270 PhaseCCP::print_statistics(); 271 PhaseRegAlloc::print_statistics(); 272 Scheduling::print_statistics(); 273 PhasePeephole::print_statistics(); 274 PhaseIdealLoop::print_statistics(); 275 if (xtty != NULL) xtty->tail("statistics"); 276 } 277 if (_intrinsic_hist_flags[vmIntrinsics::_none] != 0) { 278 // put this under its own <statistics> element. 279 print_intrinsic_statistics(); 280 } 281 } 282 #endif //PRODUCT 283 284 // Support for bundling info 285 Bundle* Compile::node_bundling(const Node *n) { 286 assert(valid_bundle_info(n), "oob"); 287 return &_node_bundling_base[n->_idx]; 288 } 289 290 bool Compile::valid_bundle_info(const Node *n) { 291 return (_node_bundling_limit > n->_idx); 292 } 293 294 295 void Compile::gvn_replace_by(Node* n, Node* nn) { 296 for (DUIterator_Last imin, i = n->last_outs(imin); i >= imin; ) { 297 Node* use = n->last_out(i); 298 bool is_in_table = initial_gvn()->hash_delete(use); 299 uint uses_found = 0; 300 for (uint j = 0; j < use->len(); j++) { 301 if (use->in(j) == n) { 302 if (j < use->req()) 303 use->set_req(j, nn); 304 else 305 use->set_prec(j, nn); 306 uses_found++; 307 } 308 } 309 if (is_in_table) { 310 // reinsert into table 311 initial_gvn()->hash_find_insert(use); 312 } 313 record_for_igvn(use); 314 i -= uses_found; // we deleted 1 or more copies of this edge 315 } 316 } 317 318 319 320 321 // Identify all nodes that are reachable from below, useful. 322 // Use breadth-first pass that records state in a Unique_Node_List, 323 // recursive traversal is slower. 324 void Compile::identify_useful_nodes(Unique_Node_List &useful) { 325 int estimated_worklist_size = unique(); 326 useful.map( estimated_worklist_size, NULL ); // preallocate space 327 328 // Initialize worklist 329 if (root() != NULL) { useful.push(root()); } 330 // If 'top' is cached, declare it useful to preserve cached node 331 if( cached_top_node() ) { useful.push(cached_top_node()); } 332 333 // Push all useful nodes onto the list, breadthfirst 334 for( uint next = 0; next < useful.size(); ++next ) { 335 assert( next < unique(), "Unique useful nodes < total nodes"); 336 Node *n = useful.at(next); 337 uint max = n->len(); 338 for( uint i = 0; i < max; ++i ) { 339 Node *m = n->in(i); 340 if( m == NULL ) continue; 341 useful.push(m); 342 } 343 } 344 } 345 346 // Disconnect all useless nodes by disconnecting those at the boundary. 347 void Compile::remove_useless_nodes(Unique_Node_List &useful) { 348 uint next = 0; 349 while( next < useful.size() ) { 350 Node *n = useful.at(next++); 351 // Use raw traversal of out edges since this code removes out edges 352 int max = n->outcnt(); 353 for (int j = 0; j < max; ++j ) { 354 Node* child = n->raw_out(j); 355 if( ! useful.member(child) ) { 356 assert( !child->is_top() || child != top(), 357 "If top is cached in Compile object it is in useful list"); 358 // Only need to remove this out-edge to the useless node 359 n->raw_del_out(j); 360 --j; 361 --max; 362 } 363 } 364 if (n->outcnt() == 1 && n->has_special_unique_user()) { 365 record_for_igvn( n->unique_out() ); 366 } 367 } 368 debug_only(verify_graph_edges(true/*check for no_dead_code*/);) 369 } 370 371 //------------------------------frame_size_in_words----------------------------- 372 // frame_slots in units of words 373 int Compile::frame_size_in_words() const { 374 // shift is 0 in LP32 and 1 in LP64 375 const int shift = (LogBytesPerWord - LogBytesPerInt); 376 int words = _frame_slots >> shift; 377 assert( words << shift == _frame_slots, "frame size must be properly aligned in LP64" ); 378 return words; 379 } 380 381 // ============================================================================ 382 //------------------------------CompileWrapper--------------------------------- 383 class CompileWrapper : public StackObj { 384 Compile *const _compile; 385 public: 386 CompileWrapper(Compile* compile); 387 388 ~CompileWrapper(); 389 }; 390 391 CompileWrapper::CompileWrapper(Compile* compile) : _compile(compile) { 392 // the Compile* pointer is stored in the current ciEnv: 393 ciEnv* env = compile->env(); 394 assert(env == ciEnv::current(), "must already be a ciEnv active"); 395 assert(env->compiler_data() == NULL, "compile already active?"); 396 env->set_compiler_data(compile); 397 assert(compile == Compile::current(), "sanity"); 398 399 compile->set_type_dict(NULL); 400 compile->set_type_hwm(NULL); 401 compile->set_type_last_size(0); 402 compile->set_last_tf(NULL, NULL); 403 compile->set_indexSet_arena(NULL); 404 compile->set_indexSet_free_block_list(NULL); 405 compile->init_type_arena(); 406 Type::Initialize(compile); 407 _compile->set_scratch_buffer_blob(NULL); 408 _compile->begin_method(); 409 } 410 CompileWrapper::~CompileWrapper() { 411 _compile->end_method(); 412 if (_compile->scratch_buffer_blob() != NULL) 413 BufferBlob::free(_compile->scratch_buffer_blob()); 414 _compile->env()->set_compiler_data(NULL); 415 } 416 417 418 //----------------------------print_compile_messages--------------------------- 419 void Compile::print_compile_messages() { 420 #ifndef PRODUCT 421 // Check if recompiling 422 if (_subsume_loads == false && PrintOpto) { 423 // Recompiling without allowing machine instructions to subsume loads 424 tty->print_cr("*********************************************************"); 425 tty->print_cr("** Bailout: Recompile without subsuming loads **"); 426 tty->print_cr("*********************************************************"); 427 } 428 if (_do_escape_analysis != DoEscapeAnalysis && PrintOpto) { 429 // Recompiling without escape analysis 430 tty->print_cr("*********************************************************"); 431 tty->print_cr("** Bailout: Recompile without escape analysis **"); 432 tty->print_cr("*********************************************************"); 433 } 434 if (env()->break_at_compile()) { 435 // Open the debugger when compiling this method. 436 tty->print("### Breaking when compiling: "); 437 method()->print_short_name(); 438 tty->cr(); 439 BREAKPOINT; 440 } 441 442 if( PrintOpto ) { 443 if (is_osr_compilation()) { 444 tty->print("[OSR]%3d", _compile_id); 445 } else { 446 tty->print("%3d", _compile_id); 447 } 448 } 449 #endif 450 } 451 452 453 //-----------------------init_scratch_buffer_blob------------------------------ 454 // Construct a temporary BufferBlob and cache it for this compile. 455 void Compile::init_scratch_buffer_blob(int const_size) { 456 // If there is already a scratch buffer blob allocated and the 457 // constant section is big enough, use it. Otherwise free the 458 // current and allocate a new one. 459 BufferBlob* blob = scratch_buffer_blob(); 460 if ((blob != NULL) && (const_size <= _scratch_const_size)) { 461 // Use the current blob. 462 } else { 463 if (blob != NULL) { 464 BufferBlob::free(blob); 465 } 466 467 ResourceMark rm; 468 _scratch_const_size = const_size; 469 int size = (MAX_inst_size + MAX_stubs_size + _scratch_const_size); 470 blob = BufferBlob::create("Compile::scratch_buffer", size); 471 // Record the buffer blob for next time. 472 set_scratch_buffer_blob(blob); 473 // Have we run out of code space? 474 if (scratch_buffer_blob() == NULL) { 475 // Let CompilerBroker disable further compilations. 476 record_failure("Not enough space for scratch buffer in CodeCache"); 477 return; 478 } 479 } 480 481 // Initialize the relocation buffers 482 relocInfo* locs_buf = (relocInfo*) blob->content_end() - MAX_locs_size; 483 set_scratch_locs_memory(locs_buf); 484 } 485 486 487 //-----------------------scratch_emit_size------------------------------------- 488 // Helper function that computes size by emitting code 489 uint Compile::scratch_emit_size(const Node* n) { 490 // Start scratch_emit_size section. 491 set_in_scratch_emit_size(true); 492 493 // Emit into a trash buffer and count bytes emitted. 494 // This is a pretty expensive way to compute a size, 495 // but it works well enough if seldom used. 496 // All common fixed-size instructions are given a size 497 // method by the AD file. 498 // Note that the scratch buffer blob and locs memory are 499 // allocated at the beginning of the compile task, and 500 // may be shared by several calls to scratch_emit_size. 501 // The allocation of the scratch buffer blob is particularly 502 // expensive, since it has to grab the code cache lock. 503 BufferBlob* blob = this->scratch_buffer_blob(); 504 assert(blob != NULL, "Initialize BufferBlob at start"); 505 assert(blob->size() > MAX_inst_size, "sanity"); 506 relocInfo* locs_buf = scratch_locs_memory(); 507 address blob_begin = blob->content_begin(); 508 address blob_end = (address)locs_buf; 509 assert(blob->content_contains(blob_end), "sanity"); 510 CodeBuffer buf(blob_begin, blob_end - blob_begin); 511 buf.initialize_consts_size(_scratch_const_size); 512 buf.initialize_stubs_size(MAX_stubs_size); 513 assert(locs_buf != NULL, "sanity"); 514 int lsize = MAX_locs_size / 3; 515 buf.consts()->initialize_shared_locs(&locs_buf[lsize * 0], lsize); 516 buf.insts()->initialize_shared_locs( &locs_buf[lsize * 1], lsize); 517 buf.stubs()->initialize_shared_locs( &locs_buf[lsize * 2], lsize); 518 519 // Do the emission. 520 n->emit(buf, this->regalloc()); 521 522 // End scratch_emit_size section. 523 set_in_scratch_emit_size(false); 524 525 return buf.insts_size(); 526 } 527 528 529 // ============================================================================ 530 //------------------------------Compile standard------------------------------- 531 debug_only( int Compile::_debug_idx = 100000; ) 532 533 // Compile a method. entry_bci is -1 for normal compilations and indicates 534 // the continuation bci for on stack replacement. 535 536 537 Compile::Compile( ciEnv* ci_env, C2Compiler* compiler, ciMethod* target, int osr_bci, bool subsume_loads, bool do_escape_analysis ) 538 : Phase(Compiler), 539 _env(ci_env), 540 _log(ci_env->log()), 541 _compile_id(ci_env->compile_id()), 542 _save_argument_registers(false), 543 _stub_name(NULL), 544 _stub_function(NULL), 545 _stub_entry_point(NULL), 546 _method(target), 547 _entry_bci(osr_bci), 548 _initial_gvn(NULL), 549 _for_igvn(NULL), 550 _warm_calls(NULL), 551 _subsume_loads(subsume_loads), 552 _do_escape_analysis(do_escape_analysis), 553 _failure_reason(NULL), 554 _code_buffer("Compile::Fill_buffer"), 555 _orig_pc_slot(0), 556 _orig_pc_slot_offset_in_bytes(0), 557 _has_method_handle_invokes(false), 558 _mach_constant_base_node(NULL), 559 _node_bundling_limit(0), 560 _node_bundling_base(NULL), 561 _java_calls(0), 562 _inner_loops(0), 563 _scratch_const_size(-1), 564 _in_scratch_emit_size(false), 565 #ifndef PRODUCT 566 _trace_opto_output(TraceOptoOutput || method()->has_option("TraceOptoOutput")), 567 _printer(IdealGraphPrinter::printer()), 568 #endif 569 _congraph(NULL) { 570 C = this; 571 572 CompileWrapper cw(this); 573 #ifndef PRODUCT 574 if (TimeCompiler2) { 575 tty->print(" "); 576 target->holder()->name()->print(); 577 tty->print("."); 578 target->print_short_name(); 579 tty->print(" "); 580 } 581 TraceTime t1("Total compilation time", &_t_totalCompilation, TimeCompiler, TimeCompiler2); 582 TraceTime t2(NULL, &_t_methodCompilation, TimeCompiler, false); 583 bool print_opto_assembly = PrintOptoAssembly || _method->has_option("PrintOptoAssembly"); 584 if (!print_opto_assembly) { 585 bool print_assembly = (PrintAssembly || _method->should_print_assembly()); 586 if (print_assembly && !Disassembler::can_decode()) { 587 tty->print_cr("PrintAssembly request changed to PrintOptoAssembly"); 588 print_opto_assembly = true; 589 } 590 } 591 set_print_assembly(print_opto_assembly); 592 set_parsed_irreducible_loop(false); 593 #endif 594 595 if (ProfileTraps) { 596 // Make sure the method being compiled gets its own MDO, 597 // so we can at least track the decompile_count(). 598 method()->ensure_method_data(); 599 } 600 601 Init(::AliasLevel); 602 603 604 print_compile_messages(); 605 606 if (UseOldInlining || PrintCompilation NOT_PRODUCT( || PrintOpto) ) 607 _ilt = InlineTree::build_inline_tree_root(); 608 else 609 _ilt = NULL; 610 611 // Even if NO memory addresses are used, MergeMem nodes must have at least 1 slice 612 assert(num_alias_types() >= AliasIdxRaw, ""); 613 614 #define MINIMUM_NODE_HASH 1023 615 // Node list that Iterative GVN will start with 616 Unique_Node_List for_igvn(comp_arena()); 617 set_for_igvn(&for_igvn); 618 619 // GVN that will be run immediately on new nodes 620 uint estimated_size = method()->code_size()*4+64; 621 estimated_size = (estimated_size < MINIMUM_NODE_HASH ? MINIMUM_NODE_HASH : estimated_size); 622 PhaseGVN gvn(node_arena(), estimated_size); 623 set_initial_gvn(&gvn); 624 625 { // Scope for timing the parser 626 TracePhase t3("parse", &_t_parser, true); 627 628 // Put top into the hash table ASAP. 629 initial_gvn()->transform_no_reclaim(top()); 630 631 // Set up tf(), start(), and find a CallGenerator. 632 CallGenerator* cg = NULL; 633 if (is_osr_compilation()) { 634 const TypeTuple *domain = StartOSRNode::osr_domain(); 635 const TypeTuple *range = TypeTuple::make_range(method()->signature()); 636 init_tf(TypeFunc::make(domain, range)); 637 StartNode* s = new (this, 2) StartOSRNode(root(), domain); 638 initial_gvn()->set_type_bottom(s); 639 init_start(s); 640 cg = CallGenerator::for_osr(method(), entry_bci()); 641 } else { 642 // Normal case. 643 init_tf(TypeFunc::make(method())); 644 StartNode* s = new (this, 2) StartNode(root(), tf()->domain()); 645 initial_gvn()->set_type_bottom(s); 646 init_start(s); 647 if (method()->intrinsic_id() == vmIntrinsics::_Reference_get && UseG1GC) { 648 // With java.lang.ref.reference.get() we must go through the 649 // intrinsic when G1 is enabled - even when get() is the root 650 // method of the compile - so that, if necessary, the value in 651 // the referent field of the reference object gets recorded by 652 // the pre-barrier code. 653 // Specifically, if G1 is enabled, the value in the referent 654 // field is recorded by the G1 SATB pre barrier. This will 655 // result in the referent being marked live and the reference 656 // object removed from the list of discovered references during 657 // reference processing. 658 cg = find_intrinsic(method(), false); 659 } 660 if (cg == NULL) { 661 float past_uses = method()->interpreter_invocation_count(); 662 float expected_uses = past_uses; 663 cg = CallGenerator::for_inline(method(), expected_uses); 664 } 665 } 666 if (failing()) return; 667 if (cg == NULL) { 668 record_method_not_compilable_all_tiers("cannot parse method"); 669 return; 670 } 671 JVMState* jvms = build_start_state(start(), tf()); 672 if ((jvms = cg->generate(jvms)) == NULL) { 673 record_method_not_compilable("method parse failed"); 674 return; 675 } 676 GraphKit kit(jvms); 677 678 if (!kit.stopped()) { 679 // Accept return values, and transfer control we know not where. 680 // This is done by a special, unique ReturnNode bound to root. 681 return_values(kit.jvms()); 682 } 683 684 if (kit.has_exceptions()) { 685 // Any exceptions that escape from this call must be rethrown 686 // to whatever caller is dynamically above us on the stack. 687 // This is done by a special, unique RethrowNode bound to root. 688 rethrow_exceptions(kit.transfer_exceptions_into_jvms()); 689 } 690 691 if (!failing() && has_stringbuilder()) { 692 { 693 // remove useless nodes to make the usage analysis simpler 694 ResourceMark rm; 695 PhaseRemoveUseless pru(initial_gvn(), &for_igvn); 696 } 697 698 { 699 ResourceMark rm; 700 print_method("Before StringOpts", 3); 701 PhaseStringOpts pso(initial_gvn(), &for_igvn); 702 print_method("After StringOpts", 3); 703 } 704 705 // now inline anything that we skipped the first time around 706 while (_late_inlines.length() > 0) { 707 CallGenerator* cg = _late_inlines.pop(); 708 cg->do_late_inline(); 709 } 710 } 711 assert(_late_inlines.length() == 0, "should have been processed"); 712 713 print_method("Before RemoveUseless", 3); 714 715 // Remove clutter produced by parsing. 716 if (!failing()) { 717 ResourceMark rm; 718 PhaseRemoveUseless pru(initial_gvn(), &for_igvn); 719 } 720 } 721 722 // Note: Large methods are capped off in do_one_bytecode(). 723 if (failing()) return; 724 725 // After parsing, node notes are no longer automagic. 726 // They must be propagated by register_new_node_with_optimizer(), 727 // clone(), or the like. 728 set_default_node_notes(NULL); 729 730 for (;;) { 731 int successes = Inline_Warm(); 732 if (failing()) return; 733 if (successes == 0) break; 734 } 735 736 // Drain the list. 737 Finish_Warm(); 738 #ifndef PRODUCT 739 if (_printer) { 740 _printer->print_inlining(this); 741 } 742 #endif 743 744 if (failing()) return; 745 NOT_PRODUCT( verify_graph_edges(); ) 746 747 // Now optimize 748 Optimize(); 749 if (failing()) return; 750 NOT_PRODUCT( verify_graph_edges(); ) 751 752 #ifndef PRODUCT 753 if (PrintIdeal) { 754 ttyLocker ttyl; // keep the following output all in one block 755 // This output goes directly to the tty, not the compiler log. 756 // To enable tools to match it up with the compilation activity, 757 // be sure to tag this tty output with the compile ID. 758 if (xtty != NULL) { 759 xtty->head("ideal compile_id='%d'%s", compile_id(), 760 is_osr_compilation() ? " compile_kind='osr'" : 761 ""); 762 } 763 root()->dump(9999); 764 if (xtty != NULL) { 765 xtty->tail("ideal"); 766 } 767 } 768 #endif 769 770 // Now that we know the size of all the monitors we can add a fixed slot 771 // for the original deopt pc. 772 773 _orig_pc_slot = fixed_slots(); 774 int next_slot = _orig_pc_slot + (sizeof(address) / VMRegImpl::stack_slot_size); 775 set_fixed_slots(next_slot); 776 777 // Now generate code 778 Code_Gen(); 779 if (failing()) return; 780 781 // Check if we want to skip execution of all compiled code. 782 { 783 #ifndef PRODUCT 784 if (OptoNoExecute) { 785 record_method_not_compilable("+OptoNoExecute"); // Flag as failed 786 return; 787 } 788 TracePhase t2("install_code", &_t_registerMethod, TimeCompiler); 789 #endif 790 791 if (is_osr_compilation()) { 792 _code_offsets.set_value(CodeOffsets::Verified_Entry, 0); 793 _code_offsets.set_value(CodeOffsets::OSR_Entry, _first_block_size); 794 } else { 795 _code_offsets.set_value(CodeOffsets::Verified_Entry, _first_block_size); 796 _code_offsets.set_value(CodeOffsets::OSR_Entry, 0); 797 } 798 799 env()->register_method(_method, _entry_bci, 800 &_code_offsets, 801 _orig_pc_slot_offset_in_bytes, 802 code_buffer(), 803 frame_size_in_words(), _oop_map_set, 804 &_handler_table, &_inc_table, 805 compiler, 806 env()->comp_level(), 807 true, /*has_debug_info*/ 808 has_unsafe_access() 809 ); 810 } 811 } 812 813 //------------------------------Compile---------------------------------------- 814 // Compile a runtime stub 815 Compile::Compile( ciEnv* ci_env, 816 TypeFunc_generator generator, 817 address stub_function, 818 const char *stub_name, 819 int is_fancy_jump, 820 bool pass_tls, 821 bool save_arg_registers, 822 bool return_pc ) 823 : Phase(Compiler), 824 _env(ci_env), 825 _log(ci_env->log()), 826 _compile_id(-1), 827 _save_argument_registers(save_arg_registers), 828 _method(NULL), 829 _stub_name(stub_name), 830 _stub_function(stub_function), 831 _stub_entry_point(NULL), 832 _entry_bci(InvocationEntryBci), 833 _initial_gvn(NULL), 834 _for_igvn(NULL), 835 _warm_calls(NULL), 836 _orig_pc_slot(0), 837 _orig_pc_slot_offset_in_bytes(0), 838 _subsume_loads(true), 839 _do_escape_analysis(false), 840 _failure_reason(NULL), 841 _code_buffer("Compile::Fill_buffer"), 842 _has_method_handle_invokes(false), 843 _mach_constant_base_node(NULL), 844 _node_bundling_limit(0), 845 _node_bundling_base(NULL), 846 _java_calls(0), 847 _inner_loops(0), 848 #ifndef PRODUCT 849 _trace_opto_output(TraceOptoOutput), 850 _printer(NULL), 851 #endif 852 _congraph(NULL) { 853 C = this; 854 855 #ifndef PRODUCT 856 TraceTime t1(NULL, &_t_totalCompilation, TimeCompiler, false); 857 TraceTime t2(NULL, &_t_stubCompilation, TimeCompiler, false); 858 set_print_assembly(PrintFrameConverterAssembly); 859 set_parsed_irreducible_loop(false); 860 #endif 861 CompileWrapper cw(this); 862 Init(/*AliasLevel=*/ 0); 863 init_tf((*generator)()); 864 865 { 866 // The following is a dummy for the sake of GraphKit::gen_stub 867 Unique_Node_List for_igvn(comp_arena()); 868 set_for_igvn(&for_igvn); // not used, but some GraphKit guys push on this 869 PhaseGVN gvn(Thread::current()->resource_area(),255); 870 set_initial_gvn(&gvn); // not significant, but GraphKit guys use it pervasively 871 gvn.transform_no_reclaim(top()); 872 873 GraphKit kit; 874 kit.gen_stub(stub_function, stub_name, is_fancy_jump, pass_tls, return_pc); 875 } 876 877 NOT_PRODUCT( verify_graph_edges(); ) 878 Code_Gen(); 879 if (failing()) return; 880 881 882 // Entry point will be accessed using compile->stub_entry_point(); 883 if (code_buffer() == NULL) { 884 Matcher::soft_match_failure(); 885 } else { 886 if (PrintAssembly && (WizardMode || Verbose)) 887 tty->print_cr("### Stub::%s", stub_name); 888 889 if (!failing()) { 890 assert(_fixed_slots == 0, "no fixed slots used for runtime stubs"); 891 892 // Make the NMethod 893 // For now we mark the frame as never safe for profile stackwalking 894 RuntimeStub *rs = RuntimeStub::new_runtime_stub(stub_name, 895 code_buffer(), 896 CodeOffsets::frame_never_safe, 897 // _code_offsets.value(CodeOffsets::Frame_Complete), 898 frame_size_in_words(), 899 _oop_map_set, 900 save_arg_registers); 901 assert(rs != NULL && rs->is_runtime_stub(), "sanity check"); 902 903 _stub_entry_point = rs->entry_point(); 904 } 905 } 906 } 907 908 #ifndef PRODUCT 909 void print_opto_verbose_signature( const TypeFunc *j_sig, const char *stub_name ) { 910 if(PrintOpto && Verbose) { 911 tty->print("%s ", stub_name); j_sig->print_flattened(); tty->cr(); 912 } 913 } 914 #endif 915 916 void Compile::print_codes() { 917 } 918 919 //------------------------------Init------------------------------------------- 920 // Prepare for a single compilation 921 void Compile::Init(int aliaslevel) { 922 _unique = 0; 923 _regalloc = NULL; 924 925 _tf = NULL; // filled in later 926 _top = NULL; // cached later 927 _matcher = NULL; // filled in later 928 _cfg = NULL; // filled in later 929 930 set_24_bit_selection_and_mode(Use24BitFP, false); 931 932 _node_note_array = NULL; 933 _default_node_notes = NULL; 934 935 _immutable_memory = NULL; // filled in at first inquiry 936 937 // Globally visible Nodes 938 // First set TOP to NULL to give safe behavior during creation of RootNode 939 set_cached_top_node(NULL); 940 set_root(new (this, 3) RootNode()); 941 // Now that you have a Root to point to, create the real TOP 942 set_cached_top_node( new (this, 1) ConNode(Type::TOP) ); 943 set_recent_alloc(NULL, NULL); 944 945 // Create Debug Information Recorder to record scopes, oopmaps, etc. 946 env()->set_oop_recorder(new OopRecorder(comp_arena())); 947 env()->set_debug_info(new DebugInformationRecorder(env()->oop_recorder())); 948 env()->set_dependencies(new Dependencies(env())); 949 950 _fixed_slots = 0; 951 set_has_split_ifs(false); 952 set_has_loops(has_method() && method()->has_loops()); // first approximation 953 set_has_stringbuilder(false); 954 _trap_can_recompile = false; // no traps emitted yet 955 _major_progress = true; // start out assuming good things will happen 956 set_has_unsafe_access(false); 957 Copy::zero_to_bytes(_trap_hist, sizeof(_trap_hist)); 958 set_decompile_count(0); 959 960 set_do_freq_based_layout(BlockLayoutByFrequency || method_has_option("BlockLayoutByFrequency")); 961 set_num_loop_opts(LoopOptsCount); 962 set_do_inlining(Inline); 963 set_max_inline_size(MaxInlineSize); 964 set_freq_inline_size(FreqInlineSize); 965 set_do_scheduling(OptoScheduling); 966 set_do_count_invocations(false); 967 set_do_method_data_update(false); 968 969 if (debug_info()->recording_non_safepoints()) { 970 set_node_note_array(new(comp_arena()) GrowableArray<Node_Notes*> 971 (comp_arena(), 8, 0, NULL)); 972 set_default_node_notes(Node_Notes::make(this)); 973 } 974 975 // // -- Initialize types before each compile -- 976 // // Update cached type information 977 // if( _method && _method->constants() ) 978 // Type::update_loaded_types(_method, _method->constants()); 979 980 // Init alias_type map. 981 if (!_do_escape_analysis && aliaslevel == 3) 982 aliaslevel = 2; // No unique types without escape analysis 983 _AliasLevel = aliaslevel; 984 const int grow_ats = 16; 985 _max_alias_types = grow_ats; 986 _alias_types = NEW_ARENA_ARRAY(comp_arena(), AliasType*, grow_ats); 987 AliasType* ats = NEW_ARENA_ARRAY(comp_arena(), AliasType, grow_ats); 988 Copy::zero_to_bytes(ats, sizeof(AliasType)*grow_ats); 989 { 990 for (int i = 0; i < grow_ats; i++) _alias_types[i] = &ats[i]; 991 } 992 // Initialize the first few types. 993 _alias_types[AliasIdxTop]->Init(AliasIdxTop, NULL); 994 _alias_types[AliasIdxBot]->Init(AliasIdxBot, TypePtr::BOTTOM); 995 _alias_types[AliasIdxRaw]->Init(AliasIdxRaw, TypeRawPtr::BOTTOM); 996 _num_alias_types = AliasIdxRaw+1; 997 // Zero out the alias type cache. 998 Copy::zero_to_bytes(_alias_cache, sizeof(_alias_cache)); 999 // A NULL adr_type hits in the cache right away. Preload the right answer. 1000 probe_alias_cache(NULL)->_index = AliasIdxTop; 1001 1002 _intrinsics = NULL; 1003 _macro_nodes = new(comp_arena()) GrowableArray<Node*>(comp_arena(), 8, 0, NULL); 1004 _predicate_opaqs = new(comp_arena()) GrowableArray<Node*>(comp_arena(), 8, 0, NULL); 1005 register_library_intrinsics(); 1006 } 1007 1008 //---------------------------init_start---------------------------------------- 1009 // Install the StartNode on this compile object. 1010 void Compile::init_start(StartNode* s) { 1011 if (failing()) 1012 return; // already failing 1013 assert(s == start(), ""); 1014 } 1015 1016 StartNode* Compile::start() const { 1017 assert(!failing(), ""); 1018 for (DUIterator_Fast imax, i = root()->fast_outs(imax); i < imax; i++) { 1019 Node* start = root()->fast_out(i); 1020 if( start->is_Start() ) 1021 return start->as_Start(); 1022 } 1023 ShouldNotReachHere(); 1024 return NULL; 1025 } 1026 1027 //-------------------------------immutable_memory------------------------------------- 1028 // Access immutable memory 1029 Node* Compile::immutable_memory() { 1030 if (_immutable_memory != NULL) { 1031 return _immutable_memory; 1032 } 1033 StartNode* s = start(); 1034 for (DUIterator_Fast imax, i = s->fast_outs(imax); true; i++) { 1035 Node *p = s->fast_out(i); 1036 if (p != s && p->as_Proj()->_con == TypeFunc::Memory) { 1037 _immutable_memory = p; 1038 return _immutable_memory; 1039 } 1040 } 1041 ShouldNotReachHere(); 1042 return NULL; 1043 } 1044 1045 //----------------------set_cached_top_node------------------------------------ 1046 // Install the cached top node, and make sure Node::is_top works correctly. 1047 void Compile::set_cached_top_node(Node* tn) { 1048 if (tn != NULL) verify_top(tn); 1049 Node* old_top = _top; 1050 _top = tn; 1051 // Calling Node::setup_is_top allows the nodes the chance to adjust 1052 // their _out arrays. 1053 if (_top != NULL) _top->setup_is_top(); 1054 if (old_top != NULL) old_top->setup_is_top(); 1055 assert(_top == NULL || top()->is_top(), ""); 1056 } 1057 1058 #ifndef PRODUCT 1059 void Compile::verify_top(Node* tn) const { 1060 if (tn != NULL) { 1061 assert(tn->is_Con(), "top node must be a constant"); 1062 assert(((ConNode*)tn)->type() == Type::TOP, "top node must have correct type"); 1063 assert(tn->in(0) != NULL, "must have live top node"); 1064 } 1065 } 1066 #endif 1067 1068 1069 ///-------------------Managing Per-Node Debug & Profile Info------------------- 1070 1071 void Compile::grow_node_notes(GrowableArray<Node_Notes*>* arr, int grow_by) { 1072 guarantee(arr != NULL, ""); 1073 int num_blocks = arr->length(); 1074 if (grow_by < num_blocks) grow_by = num_blocks; 1075 int num_notes = grow_by * _node_notes_block_size; 1076 Node_Notes* notes = NEW_ARENA_ARRAY(node_arena(), Node_Notes, num_notes); 1077 Copy::zero_to_bytes(notes, num_notes * sizeof(Node_Notes)); 1078 while (num_notes > 0) { 1079 arr->append(notes); 1080 notes += _node_notes_block_size; 1081 num_notes -= _node_notes_block_size; 1082 } 1083 assert(num_notes == 0, "exact multiple, please"); 1084 } 1085 1086 bool Compile::copy_node_notes_to(Node* dest, Node* source) { 1087 if (source == NULL || dest == NULL) return false; 1088 1089 if (dest->is_Con()) 1090 return false; // Do not push debug info onto constants. 1091 1092 #ifdef ASSERT 1093 // Leave a bread crumb trail pointing to the original node: 1094 if (dest != NULL && dest != source && dest->debug_orig() == NULL) { 1095 dest->set_debug_orig(source); 1096 } 1097 #endif 1098 1099 if (node_note_array() == NULL) 1100 return false; // Not collecting any notes now. 1101 1102 // This is a copy onto a pre-existing node, which may already have notes. 1103 // If both nodes have notes, do not overwrite any pre-existing notes. 1104 Node_Notes* source_notes = node_notes_at(source->_idx); 1105 if (source_notes == NULL || source_notes->is_clear()) return false; 1106 Node_Notes* dest_notes = node_notes_at(dest->_idx); 1107 if (dest_notes == NULL || dest_notes->is_clear()) { 1108 return set_node_notes_at(dest->_idx, source_notes); 1109 } 1110 1111 Node_Notes merged_notes = (*source_notes); 1112 // The order of operations here ensures that dest notes will win... 1113 merged_notes.update_from(dest_notes); 1114 return set_node_notes_at(dest->_idx, &merged_notes); 1115 } 1116 1117 1118 //--------------------------allow_range_check_smearing------------------------- 1119 // Gating condition for coalescing similar range checks. 1120 // Sometimes we try 'speculatively' replacing a series of a range checks by a 1121 // single covering check that is at least as strong as any of them. 1122 // If the optimization succeeds, the simplified (strengthened) range check 1123 // will always succeed. If it fails, we will deopt, and then give up 1124 // on the optimization. 1125 bool Compile::allow_range_check_smearing() const { 1126 // If this method has already thrown a range-check, 1127 // assume it was because we already tried range smearing 1128 // and it failed. 1129 uint already_trapped = trap_count(Deoptimization::Reason_range_check); 1130 return !already_trapped; 1131 } 1132 1133 1134 //------------------------------flatten_alias_type----------------------------- 1135 const TypePtr *Compile::flatten_alias_type( const TypePtr *tj ) const { 1136 int offset = tj->offset(); 1137 TypePtr::PTR ptr = tj->ptr(); 1138 1139 // Known instance (scalarizable allocation) alias only with itself. 1140 bool is_known_inst = tj->isa_oopptr() != NULL && 1141 tj->is_oopptr()->is_known_instance(); 1142 1143 // Process weird unsafe references. 1144 if (offset == Type::OffsetBot && (tj->isa_instptr() /*|| tj->isa_klassptr()*/)) { 1145 assert(InlineUnsafeOps, "indeterminate pointers come only from unsafe ops"); 1146 assert(!is_known_inst, "scalarizable allocation should not have unsafe references"); 1147 tj = TypeOopPtr::BOTTOM; 1148 ptr = tj->ptr(); 1149 offset = tj->offset(); 1150 } 1151 1152 // Array pointers need some flattening 1153 const TypeAryPtr *ta = tj->isa_aryptr(); 1154 if( ta && is_known_inst ) { 1155 if ( offset != Type::OffsetBot && 1156 offset > arrayOopDesc::length_offset_in_bytes() ) { 1157 offset = Type::OffsetBot; // Flatten constant access into array body only 1158 tj = ta = TypeAryPtr::make(ptr, ta->ary(), ta->klass(), true, offset, ta->instance_id()); 1159 } 1160 } else if( ta && _AliasLevel >= 2 ) { 1161 // For arrays indexed by constant indices, we flatten the alias 1162 // space to include all of the array body. Only the header, klass 1163 // and array length can be accessed un-aliased. 1164 if( offset != Type::OffsetBot ) { 1165 if( ta->const_oop() ) { // methodDataOop or methodOop 1166 offset = Type::OffsetBot; // Flatten constant access into array body 1167 tj = ta = TypeAryPtr::make(ptr,ta->const_oop(),ta->ary(),ta->klass(),false,offset); 1168 } else if( offset == arrayOopDesc::length_offset_in_bytes() ) { 1169 // range is OK as-is. 1170 tj = ta = TypeAryPtr::RANGE; 1171 } else if( offset == oopDesc::klass_offset_in_bytes() ) { 1172 tj = TypeInstPtr::KLASS; // all klass loads look alike 1173 ta = TypeAryPtr::RANGE; // generic ignored junk 1174 ptr = TypePtr::BotPTR; 1175 } else if( offset == oopDesc::mark_offset_in_bytes() ) { 1176 tj = TypeInstPtr::MARK; 1177 ta = TypeAryPtr::RANGE; // generic ignored junk 1178 ptr = TypePtr::BotPTR; 1179 } else { // Random constant offset into array body 1180 offset = Type::OffsetBot; // Flatten constant access into array body 1181 tj = ta = TypeAryPtr::make(ptr,ta->ary(),ta->klass(),false,offset); 1182 } 1183 } 1184 // Arrays of fixed size alias with arrays of unknown size. 1185 if (ta->size() != TypeInt::POS) { 1186 const TypeAry *tary = TypeAry::make(ta->elem(), TypeInt::POS); 1187 tj = ta = TypeAryPtr::make(ptr,ta->const_oop(),tary,ta->klass(),false,offset); 1188 } 1189 // Arrays of known objects become arrays of unknown objects. 1190 if (ta->elem()->isa_narrowoop() && ta->elem() != TypeNarrowOop::BOTTOM) { 1191 const TypeAry *tary = TypeAry::make(TypeNarrowOop::BOTTOM, ta->size()); 1192 tj = ta = TypeAryPtr::make(ptr,ta->const_oop(),tary,NULL,false,offset); 1193 } 1194 if (ta->elem()->isa_oopptr() && ta->elem() != TypeInstPtr::BOTTOM) { 1195 const TypeAry *tary = TypeAry::make(TypeInstPtr::BOTTOM, ta->size()); 1196 tj = ta = TypeAryPtr::make(ptr,ta->const_oop(),tary,NULL,false,offset); 1197 } 1198 // Arrays of bytes and of booleans both use 'bastore' and 'baload' so 1199 // cannot be distinguished by bytecode alone. 1200 if (ta->elem() == TypeInt::BOOL) { 1201 const TypeAry *tary = TypeAry::make(TypeInt::BYTE, ta->size()); 1202 ciKlass* aklass = ciTypeArrayKlass::make(T_BYTE); 1203 tj = ta = TypeAryPtr::make(ptr,ta->const_oop(),tary,aklass,false,offset); 1204 } 1205 // During the 2nd round of IterGVN, NotNull castings are removed. 1206 // Make sure the Bottom and NotNull variants alias the same. 1207 // Also, make sure exact and non-exact variants alias the same. 1208 if( ptr == TypePtr::NotNull || ta->klass_is_exact() ) { 1209 if (ta->const_oop()) { 1210 tj = ta = TypeAryPtr::make(TypePtr::Constant,ta->const_oop(),ta->ary(),ta->klass(),false,offset); 1211 } else { 1212 tj = ta = TypeAryPtr::make(TypePtr::BotPTR,ta->ary(),ta->klass(),false,offset); 1213 } 1214 } 1215 } 1216 1217 // Oop pointers need some flattening 1218 const TypeInstPtr *to = tj->isa_instptr(); 1219 if( to && _AliasLevel >= 2 && to != TypeOopPtr::BOTTOM ) { 1220 ciInstanceKlass *k = to->klass()->as_instance_klass(); 1221 if( ptr == TypePtr::Constant ) { 1222 if (to->klass() != ciEnv::current()->Class_klass() || 1223 offset < k->size_helper() * wordSize) { 1224 // No constant oop pointers (such as Strings); they alias with 1225 // unknown strings. 1226 assert(!is_known_inst, "not scalarizable allocation"); 1227 tj = to = TypeInstPtr::make(TypePtr::BotPTR,to->klass(),false,0,offset); 1228 } 1229 } else if( is_known_inst ) { 1230 tj = to; // Keep NotNull and klass_is_exact for instance type 1231 } else if( ptr == TypePtr::NotNull || to->klass_is_exact() ) { 1232 // During the 2nd round of IterGVN, NotNull castings are removed. 1233 // Make sure the Bottom and NotNull variants alias the same. 1234 // Also, make sure exact and non-exact variants alias the same. 1235 tj = to = TypeInstPtr::make(TypePtr::BotPTR,to->klass(),false,0,offset); 1236 } 1237 // Canonicalize the holder of this field 1238 if (offset >= 0 && offset < instanceOopDesc::base_offset_in_bytes()) { 1239 // First handle header references such as a LoadKlassNode, even if the 1240 // object's klass is unloaded at compile time (4965979). 1241 if (!is_known_inst) { // Do it only for non-instance types 1242 tj = to = TypeInstPtr::make(TypePtr::BotPTR, env()->Object_klass(), false, NULL, offset); 1243 } 1244 } else if (offset < 0 || offset >= k->size_helper() * wordSize) { 1245 // Static fields are in the space above the normal instance 1246 // fields in the java.lang.Class instance. 1247 if (to->klass() != ciEnv::current()->Class_klass()) { 1248 to = NULL; 1249 tj = TypeOopPtr::BOTTOM; 1250 offset = tj->offset(); 1251 } 1252 } else { 1253 ciInstanceKlass *canonical_holder = k->get_canonical_holder(offset); 1254 if (!k->equals(canonical_holder) || tj->offset() != offset) { 1255 if( is_known_inst ) { 1256 tj = to = TypeInstPtr::make(to->ptr(), canonical_holder, true, NULL, offset, to->instance_id()); 1257 } else { 1258 tj = to = TypeInstPtr::make(to->ptr(), canonical_holder, false, NULL, offset); 1259 } 1260 } 1261 } 1262 } 1263 1264 // Klass pointers to object array klasses need some flattening 1265 const TypeKlassPtr *tk = tj->isa_klassptr(); 1266 if( tk ) { 1267 // If we are referencing a field within a Klass, we need 1268 // to assume the worst case of an Object. Both exact and 1269 // inexact types must flatten to the same alias class. 1270 // Since the flattened result for a klass is defined to be 1271 // precisely java.lang.Object, use a constant ptr. 1272 if ( offset == Type::OffsetBot || (offset >= 0 && (size_t)offset < sizeof(Klass)) ) { 1273 1274 tj = tk = TypeKlassPtr::make(TypePtr::Constant, 1275 TypeKlassPtr::OBJECT->klass(), 1276 offset); 1277 } 1278 1279 ciKlass* klass = tk->klass(); 1280 if( klass->is_obj_array_klass() ) { 1281 ciKlass* k = TypeAryPtr::OOPS->klass(); 1282 if( !k || !k->is_loaded() ) // Only fails for some -Xcomp runs 1283 k = TypeInstPtr::BOTTOM->klass(); 1284 tj = tk = TypeKlassPtr::make( TypePtr::NotNull, k, offset ); 1285 } 1286 1287 // Check for precise loads from the primary supertype array and force them 1288 // to the supertype cache alias index. Check for generic array loads from 1289 // the primary supertype array and also force them to the supertype cache 1290 // alias index. Since the same load can reach both, we need to merge 1291 // these 2 disparate memories into the same alias class. Since the 1292 // primary supertype array is read-only, there's no chance of confusion 1293 // where we bypass an array load and an array store. 1294 uint off2 = offset - Klass::primary_supers_offset_in_bytes(); 1295 if( offset == Type::OffsetBot || 1296 off2 < Klass::primary_super_limit()*wordSize ) { 1297 offset = sizeof(oopDesc) +Klass::secondary_super_cache_offset_in_bytes(); 1298 tj = tk = TypeKlassPtr::make( TypePtr::NotNull, tk->klass(), offset ); 1299 } 1300 } 1301 1302 // Flatten all Raw pointers together. 1303 if (tj->base() == Type::RawPtr) 1304 tj = TypeRawPtr::BOTTOM; 1305 1306 if (tj->base() == Type::AnyPtr) 1307 tj = TypePtr::BOTTOM; // An error, which the caller must check for. 1308 1309 // Flatten all to bottom for now 1310 switch( _AliasLevel ) { 1311 case 0: 1312 tj = TypePtr::BOTTOM; 1313 break; 1314 case 1: // Flatten to: oop, static, field or array 1315 switch (tj->base()) { 1316 //case Type::AryPtr: tj = TypeAryPtr::RANGE; break; 1317 case Type::RawPtr: tj = TypeRawPtr::BOTTOM; break; 1318 case Type::AryPtr: // do not distinguish arrays at all 1319 case Type::InstPtr: tj = TypeInstPtr::BOTTOM; break; 1320 case Type::KlassPtr: tj = TypeKlassPtr::OBJECT; break; 1321 case Type::AnyPtr: tj = TypePtr::BOTTOM; break; // caller checks it 1322 default: ShouldNotReachHere(); 1323 } 1324 break; 1325 case 2: // No collapsing at level 2; keep all splits 1326 case 3: // No collapsing at level 3; keep all splits 1327 break; 1328 default: 1329 Unimplemented(); 1330 } 1331 1332 offset = tj->offset(); 1333 assert( offset != Type::OffsetTop, "Offset has fallen from constant" ); 1334 1335 assert( (offset != Type::OffsetBot && tj->base() != Type::AryPtr) || 1336 (offset == Type::OffsetBot && tj->base() == Type::AryPtr) || 1337 (offset == Type::OffsetBot && tj == TypeOopPtr::BOTTOM) || 1338 (offset == Type::OffsetBot && tj == TypePtr::BOTTOM) || 1339 (offset == oopDesc::mark_offset_in_bytes() && tj->base() == Type::AryPtr) || 1340 (offset == oopDesc::klass_offset_in_bytes() && tj->base() == Type::AryPtr) || 1341 (offset == arrayOopDesc::length_offset_in_bytes() && tj->base() == Type::AryPtr) , 1342 "For oops, klasses, raw offset must be constant; for arrays the offset is never known" ); 1343 assert( tj->ptr() != TypePtr::TopPTR && 1344 tj->ptr() != TypePtr::AnyNull && 1345 tj->ptr() != TypePtr::Null, "No imprecise addresses" ); 1346 // assert( tj->ptr() != TypePtr::Constant || 1347 // tj->base() == Type::RawPtr || 1348 // tj->base() == Type::KlassPtr, "No constant oop addresses" ); 1349 1350 return tj; 1351 } 1352 1353 void Compile::AliasType::Init(int i, const TypePtr* at) { 1354 _index = i; 1355 _adr_type = at; 1356 _field = NULL; 1357 _is_rewritable = true; // default 1358 const TypeOopPtr *atoop = (at != NULL) ? at->isa_oopptr() : NULL; 1359 if (atoop != NULL && atoop->is_known_instance()) { 1360 const TypeOopPtr *gt = atoop->cast_to_instance_id(TypeOopPtr::InstanceBot); 1361 _general_index = Compile::current()->get_alias_index(gt); 1362 } else { 1363 _general_index = 0; 1364 } 1365 } 1366 1367 //---------------------------------print_on------------------------------------ 1368 #ifndef PRODUCT 1369 void Compile::AliasType::print_on(outputStream* st) { 1370 if (index() < 10) 1371 st->print("@ <%d> ", index()); 1372 else st->print("@ <%d>", index()); 1373 st->print(is_rewritable() ? " " : " RO"); 1374 int offset = adr_type()->offset(); 1375 if (offset == Type::OffsetBot) 1376 st->print(" +any"); 1377 else st->print(" +%-3d", offset); 1378 st->print(" in "); 1379 adr_type()->dump_on(st); 1380 const TypeOopPtr* tjp = adr_type()->isa_oopptr(); 1381 if (field() != NULL && tjp) { 1382 if (tjp->klass() != field()->holder() || 1383 tjp->offset() != field()->offset_in_bytes()) { 1384 st->print(" != "); 1385 field()->print(); 1386 st->print(" ***"); 1387 } 1388 } 1389 } 1390 1391 void print_alias_types() { 1392 Compile* C = Compile::current(); 1393 tty->print_cr("--- Alias types, AliasIdxBot .. %d", C->num_alias_types()-1); 1394 for (int idx = Compile::AliasIdxBot; idx < C->num_alias_types(); idx++) { 1395 C->alias_type(idx)->print_on(tty); 1396 tty->cr(); 1397 } 1398 } 1399 #endif 1400 1401 1402 //----------------------------probe_alias_cache-------------------------------- 1403 Compile::AliasCacheEntry* Compile::probe_alias_cache(const TypePtr* adr_type) { 1404 intptr_t key = (intptr_t) adr_type; 1405 key ^= key >> logAliasCacheSize; 1406 return &_alias_cache[key & right_n_bits(logAliasCacheSize)]; 1407 } 1408 1409 1410 //-----------------------------grow_alias_types-------------------------------- 1411 void Compile::grow_alias_types() { 1412 const int old_ats = _max_alias_types; // how many before? 1413 const int new_ats = old_ats; // how many more? 1414 const int grow_ats = old_ats+new_ats; // how many now? 1415 _max_alias_types = grow_ats; 1416 _alias_types = REALLOC_ARENA_ARRAY(comp_arena(), AliasType*, _alias_types, old_ats, grow_ats); 1417 AliasType* ats = NEW_ARENA_ARRAY(comp_arena(), AliasType, new_ats); 1418 Copy::zero_to_bytes(ats, sizeof(AliasType)*new_ats); 1419 for (int i = 0; i < new_ats; i++) _alias_types[old_ats+i] = &ats[i]; 1420 } 1421 1422 1423 //--------------------------------find_alias_type------------------------------ 1424 Compile::AliasType* Compile::find_alias_type(const TypePtr* adr_type, bool no_create, ciField* original_field) { 1425 if (_AliasLevel == 0) 1426 return alias_type(AliasIdxBot); 1427 1428 AliasCacheEntry* ace = probe_alias_cache(adr_type); 1429 if (ace->_adr_type == adr_type) { 1430 return alias_type(ace->_index); 1431 } 1432 1433 // Handle special cases. 1434 if (adr_type == NULL) return alias_type(AliasIdxTop); 1435 if (adr_type == TypePtr::BOTTOM) return alias_type(AliasIdxBot); 1436 1437 // Do it the slow way. 1438 const TypePtr* flat = flatten_alias_type(adr_type); 1439 1440 #ifdef ASSERT 1441 assert(flat == flatten_alias_type(flat), "idempotent"); 1442 assert(flat != TypePtr::BOTTOM, "cannot alias-analyze an untyped ptr"); 1443 if (flat->isa_oopptr() && !flat->isa_klassptr()) { 1444 const TypeOopPtr* foop = flat->is_oopptr(); 1445 // Scalarizable allocations have exact klass always. 1446 bool exact = !foop->klass_is_exact() || foop->is_known_instance(); 1447 const TypePtr* xoop = foop->cast_to_exactness(exact)->is_ptr(); 1448 assert(foop == flatten_alias_type(xoop), "exactness must not affect alias type"); 1449 } 1450 assert(flat == flatten_alias_type(flat), "exact bit doesn't matter"); 1451 #endif 1452 1453 int idx = AliasIdxTop; 1454 for (int i = 0; i < num_alias_types(); i++) { 1455 if (alias_type(i)->adr_type() == flat) { 1456 idx = i; 1457 break; 1458 } 1459 } 1460 1461 if (idx == AliasIdxTop) { 1462 if (no_create) return NULL; 1463 // Grow the array if necessary. 1464 if (_num_alias_types == _max_alias_types) grow_alias_types(); 1465 // Add a new alias type. 1466 idx = _num_alias_types++; 1467 _alias_types[idx]->Init(idx, flat); 1468 if (flat == TypeInstPtr::KLASS) alias_type(idx)->set_rewritable(false); 1469 if (flat == TypeAryPtr::RANGE) alias_type(idx)->set_rewritable(false); 1470 if (flat->isa_instptr()) { 1471 if (flat->offset() == java_lang_Class::klass_offset_in_bytes() 1472 && flat->is_instptr()->klass() == env()->Class_klass()) 1473 alias_type(idx)->set_rewritable(false); 1474 } 1475 if (flat->isa_klassptr()) { 1476 if (flat->offset() == Klass::super_check_offset_offset_in_bytes() + (int)sizeof(oopDesc)) 1477 alias_type(idx)->set_rewritable(false); 1478 if (flat->offset() == Klass::modifier_flags_offset_in_bytes() + (int)sizeof(oopDesc)) 1479 alias_type(idx)->set_rewritable(false); 1480 if (flat->offset() == Klass::access_flags_offset_in_bytes() + (int)sizeof(oopDesc)) 1481 alias_type(idx)->set_rewritable(false); 1482 if (flat->offset() == Klass::java_mirror_offset_in_bytes() + (int)sizeof(oopDesc)) 1483 alias_type(idx)->set_rewritable(false); 1484 } 1485 // %%% (We would like to finalize JavaThread::threadObj_offset(), 1486 // but the base pointer type is not distinctive enough to identify 1487 // references into JavaThread.) 1488 1489 // Check for final fields. 1490 const TypeInstPtr* tinst = flat->isa_instptr(); 1491 if (tinst && tinst->offset() >= instanceOopDesc::base_offset_in_bytes()) { 1492 ciField* field; 1493 if (tinst->const_oop() != NULL && 1494 tinst->klass() == ciEnv::current()->Class_klass() && 1495 tinst->offset() >= (tinst->klass()->as_instance_klass()->size_helper() * wordSize)) { 1496 // static field 1497 ciInstanceKlass* k = tinst->const_oop()->as_instance()->java_lang_Class_klass()->as_instance_klass(); 1498 field = k->get_field_by_offset(tinst->offset(), true); 1499 } else { 1500 ciInstanceKlass *k = tinst->klass()->as_instance_klass(); 1501 field = k->get_field_by_offset(tinst->offset(), false); 1502 } 1503 assert(field == NULL || 1504 original_field == NULL || 1505 (field->holder() == original_field->holder() && 1506 field->offset() == original_field->offset() && 1507 field->is_static() == original_field->is_static()), "wrong field?"); 1508 // Set field() and is_rewritable() attributes. 1509 if (field != NULL) alias_type(idx)->set_field(field); 1510 } 1511 } 1512 1513 // Fill the cache for next time. 1514 ace->_adr_type = adr_type; 1515 ace->_index = idx; 1516 assert(alias_type(adr_type) == alias_type(idx), "type must be installed"); 1517 1518 // Might as well try to fill the cache for the flattened version, too. 1519 AliasCacheEntry* face = probe_alias_cache(flat); 1520 if (face->_adr_type == NULL) { 1521 face->_adr_type = flat; 1522 face->_index = idx; 1523 assert(alias_type(flat) == alias_type(idx), "flat type must work too"); 1524 } 1525 1526 return alias_type(idx); 1527 } 1528 1529 1530 Compile::AliasType* Compile::alias_type(ciField* field) { 1531 const TypeOopPtr* t; 1532 if (field->is_static()) 1533 t = TypeInstPtr::make(field->holder()->java_mirror()); 1534 else 1535 t = TypeOopPtr::make_from_klass_raw(field->holder()); 1536 AliasType* atp = alias_type(t->add_offset(field->offset_in_bytes()), field); 1537 assert(field->is_final() == !atp->is_rewritable(), "must get the rewritable bits correct"); 1538 return atp; 1539 } 1540 1541 1542 //------------------------------have_alias_type-------------------------------- 1543 bool Compile::have_alias_type(const TypePtr* adr_type) { 1544 AliasCacheEntry* ace = probe_alias_cache(adr_type); 1545 if (ace->_adr_type == adr_type) { 1546 return true; 1547 } 1548 1549 // Handle special cases. 1550 if (adr_type == NULL) return true; 1551 if (adr_type == TypePtr::BOTTOM) return true; 1552 1553 return find_alias_type(adr_type, true, NULL) != NULL; 1554 } 1555 1556 //-----------------------------must_alias-------------------------------------- 1557 // True if all values of the given address type are in the given alias category. 1558 bool Compile::must_alias(const TypePtr* adr_type, int alias_idx) { 1559 if (alias_idx == AliasIdxBot) return true; // the universal category 1560 if (adr_type == NULL) return true; // NULL serves as TypePtr::TOP 1561 if (alias_idx == AliasIdxTop) return false; // the empty category 1562 if (adr_type->base() == Type::AnyPtr) return false; // TypePtr::BOTTOM or its twins 1563 1564 // the only remaining possible overlap is identity 1565 int adr_idx = get_alias_index(adr_type); 1566 assert(adr_idx != AliasIdxBot && adr_idx != AliasIdxTop, ""); 1567 assert(adr_idx == alias_idx || 1568 (alias_type(alias_idx)->adr_type() != TypeOopPtr::BOTTOM 1569 && adr_type != TypeOopPtr::BOTTOM), 1570 "should not be testing for overlap with an unsafe pointer"); 1571 return adr_idx == alias_idx; 1572 } 1573 1574 //------------------------------can_alias-------------------------------------- 1575 // True if any values of the given address type are in the given alias category. 1576 bool Compile::can_alias(const TypePtr* adr_type, int alias_idx) { 1577 if (alias_idx == AliasIdxTop) return false; // the empty category 1578 if (adr_type == NULL) return false; // NULL serves as TypePtr::TOP 1579 if (alias_idx == AliasIdxBot) return true; // the universal category 1580 if (adr_type->base() == Type::AnyPtr) return true; // TypePtr::BOTTOM or its twins 1581 1582 // the only remaining possible overlap is identity 1583 int adr_idx = get_alias_index(adr_type); 1584 assert(adr_idx != AliasIdxBot && adr_idx != AliasIdxTop, ""); 1585 return adr_idx == alias_idx; 1586 } 1587 1588 1589 1590 //---------------------------pop_warm_call------------------------------------- 1591 WarmCallInfo* Compile::pop_warm_call() { 1592 WarmCallInfo* wci = _warm_calls; 1593 if (wci != NULL) _warm_calls = wci->remove_from(wci); 1594 return wci; 1595 } 1596 1597 //----------------------------Inline_Warm-------------------------------------- 1598 int Compile::Inline_Warm() { 1599 // If there is room, try to inline some more warm call sites. 1600 // %%% Do a graph index compaction pass when we think we're out of space? 1601 if (!InlineWarmCalls) return 0; 1602 1603 int calls_made_hot = 0; 1604 int room_to_grow = NodeCountInliningCutoff - unique(); 1605 int amount_to_grow = MIN2(room_to_grow, (int)NodeCountInliningStep); 1606 int amount_grown = 0; 1607 WarmCallInfo* call; 1608 while (amount_to_grow > 0 && (call = pop_warm_call()) != NULL) { 1609 int est_size = (int)call->size(); 1610 if (est_size > (room_to_grow - amount_grown)) { 1611 // This one won't fit anyway. Get rid of it. 1612 call->make_cold(); 1613 continue; 1614 } 1615 call->make_hot(); 1616 calls_made_hot++; 1617 amount_grown += est_size; 1618 amount_to_grow -= est_size; 1619 } 1620 1621 if (calls_made_hot > 0) set_major_progress(); 1622 return calls_made_hot; 1623 } 1624 1625 1626 //----------------------------Finish_Warm-------------------------------------- 1627 void Compile::Finish_Warm() { 1628 if (!InlineWarmCalls) return; 1629 if (failing()) return; 1630 if (warm_calls() == NULL) return; 1631 1632 // Clean up loose ends, if we are out of space for inlining. 1633 WarmCallInfo* call; 1634 while ((call = pop_warm_call()) != NULL) { 1635 call->make_cold(); 1636 } 1637 } 1638 1639 //---------------------cleanup_loop_predicates----------------------- 1640 // Remove the opaque nodes that protect the predicates so that all unused 1641 // checks and uncommon_traps will be eliminated from the ideal graph 1642 void Compile::cleanup_loop_predicates(PhaseIterGVN &igvn) { 1643 if (predicate_count()==0) return; 1644 for (int i = predicate_count(); i > 0; i--) { 1645 Node * n = predicate_opaque1_node(i-1); 1646 assert(n->Opcode() == Op_Opaque1, "must be"); 1647 igvn.replace_node(n, n->in(1)); 1648 } 1649 assert(predicate_count()==0, "should be clean!"); 1650 igvn.optimize(); 1651 } 1652 1653 //------------------------------Optimize--------------------------------------- 1654 // Given a graph, optimize it. 1655 void Compile::Optimize() { 1656 TracePhase t1("optimizer", &_t_optimizer, true); 1657 1658 #ifndef PRODUCT 1659 if (env()->break_at_compile()) { 1660 BREAKPOINT; 1661 } 1662 1663 #endif 1664 1665 ResourceMark rm; 1666 int loop_opts_cnt; 1667 1668 NOT_PRODUCT( verify_graph_edges(); ) 1669 1670 print_method("After Parsing"); 1671 1672 { 1673 // Iterative Global Value Numbering, including ideal transforms 1674 // Initialize IterGVN with types and values from parse-time GVN 1675 PhaseIterGVN igvn(initial_gvn()); 1676 { 1677 NOT_PRODUCT( TracePhase t2("iterGVN", &_t_iterGVN, TimeCompiler); ) 1678 igvn.optimize(); 1679 } 1680 1681 print_method("Iter GVN 1", 2); 1682 1683 if (failing()) return; 1684 1685 // Perform escape analysis 1686 if (_do_escape_analysis && ConnectionGraph::has_candidates(this)) { 1687 TracePhase t2("escapeAnalysis", &_t_escapeAnalysis, true); 1688 ConnectionGraph::do_analysis(this, &igvn); 1689 1690 if (failing()) return; 1691 1692 igvn.optimize(); 1693 print_method("Iter GVN 3", 2); 1694 1695 if (failing()) return; 1696 1697 } 1698 1699 // Loop transforms on the ideal graph. Range Check Elimination, 1700 // peeling, unrolling, etc. 1701 1702 // Set loop opts counter 1703 loop_opts_cnt = num_loop_opts(); 1704 if((loop_opts_cnt > 0) && (has_loops() || has_split_ifs())) { 1705 { 1706 TracePhase t2("idealLoop", &_t_idealLoop, true); 1707 PhaseIdealLoop ideal_loop( igvn, true, UseLoopPredicate); 1708 loop_opts_cnt--; 1709 if (major_progress()) print_method("PhaseIdealLoop 1", 2); 1710 if (failing()) return; 1711 } 1712 // Loop opts pass if partial peeling occurred in previous pass 1713 if(PartialPeelLoop && major_progress() && (loop_opts_cnt > 0)) { 1714 TracePhase t3("idealLoop", &_t_idealLoop, true); 1715 PhaseIdealLoop ideal_loop( igvn, false, UseLoopPredicate); 1716 loop_opts_cnt--; 1717 if (major_progress()) print_method("PhaseIdealLoop 2", 2); 1718 if (failing()) return; 1719 } 1720 // Loop opts pass for loop-unrolling before CCP 1721 if(major_progress() && (loop_opts_cnt > 0)) { 1722 TracePhase t4("idealLoop", &_t_idealLoop, true); 1723 PhaseIdealLoop ideal_loop( igvn, false, UseLoopPredicate); 1724 loop_opts_cnt--; 1725 if (major_progress()) print_method("PhaseIdealLoop 3", 2); 1726 } 1727 if (!failing()) { 1728 // Verify that last round of loop opts produced a valid graph 1729 NOT_PRODUCT( TracePhase t2("idealLoopVerify", &_t_idealLoopVerify, TimeCompiler); ) 1730 PhaseIdealLoop::verify(igvn); 1731 } 1732 } 1733 if (failing()) return; 1734 1735 // Conditional Constant Propagation; 1736 PhaseCCP ccp( &igvn ); 1737 assert( true, "Break here to ccp.dump_nodes_and_types(_root,999,1)"); 1738 { 1739 TracePhase t2("ccp", &_t_ccp, true); 1740 ccp.do_transform(); 1741 } 1742 print_method("PhaseCPP 1", 2); 1743 1744 assert( true, "Break here to ccp.dump_old2new_map()"); 1745 1746 // Iterative Global Value Numbering, including ideal transforms 1747 { 1748 NOT_PRODUCT( TracePhase t2("iterGVN2", &_t_iterGVN2, TimeCompiler); ) 1749 igvn = ccp; 1750 igvn.optimize(); 1751 } 1752 1753 print_method("Iter GVN 2", 2); 1754 1755 if (failing()) return; 1756 1757 // Loop transforms on the ideal graph. Range Check Elimination, 1758 // peeling, unrolling, etc. 1759 if(loop_opts_cnt > 0) { 1760 debug_only( int cnt = 0; ); 1761 bool loop_predication = UseLoopPredicate; 1762 while(major_progress() && (loop_opts_cnt > 0)) { 1763 TracePhase t2("idealLoop", &_t_idealLoop, true); 1764 assert( cnt++ < 40, "infinite cycle in loop optimization" ); 1765 PhaseIdealLoop ideal_loop( igvn, true, loop_predication); 1766 loop_opts_cnt--; 1767 if (major_progress()) print_method("PhaseIdealLoop iterations", 2); 1768 if (failing()) return; 1769 // Perform loop predication optimization during first iteration after CCP. 1770 // After that switch it off and cleanup unused loop predicates. 1771 if (loop_predication) { 1772 loop_predication = false; 1773 cleanup_loop_predicates(igvn); 1774 if (failing()) return; 1775 } 1776 } 1777 } 1778 1779 { 1780 // Verify that all previous optimizations produced a valid graph 1781 // at least to this point, even if no loop optimizations were done. 1782 NOT_PRODUCT( TracePhase t2("idealLoopVerify", &_t_idealLoopVerify, TimeCompiler); ) 1783 PhaseIdealLoop::verify(igvn); 1784 } 1785 1786 { 1787 NOT_PRODUCT( TracePhase t2("macroExpand", &_t_macroExpand, TimeCompiler); ) 1788 PhaseMacroExpand mex(igvn); 1789 if (mex.expand_macro_nodes()) { 1790 assert(failing(), "must bail out w/ explicit message"); 1791 return; 1792 } 1793 } 1794 1795 } // (End scope of igvn; run destructor if necessary for asserts.) 1796 1797 // A method with only infinite loops has no edges entering loops from root 1798 { 1799 NOT_PRODUCT( TracePhase t2("graphReshape", &_t_graphReshaping, TimeCompiler); ) 1800 if (final_graph_reshaping()) { 1801 assert(failing(), "must bail out w/ explicit message"); 1802 return; 1803 } 1804 } 1805 1806 print_method("Optimize finished", 2); 1807 } 1808 1809 1810 //------------------------------Code_Gen--------------------------------------- 1811 // Given a graph, generate code for it 1812 void Compile::Code_Gen() { 1813 if (failing()) return; 1814 1815 // Perform instruction selection. You might think we could reclaim Matcher 1816 // memory PDQ, but actually the Matcher is used in generating spill code. 1817 // Internals of the Matcher (including some VectorSets) must remain live 1818 // for awhile - thus I cannot reclaim Matcher memory lest a VectorSet usage 1819 // set a bit in reclaimed memory. 1820 1821 // In debug mode can dump m._nodes.dump() for mapping of ideal to machine 1822 // nodes. Mapping is only valid at the root of each matched subtree. 1823 NOT_PRODUCT( verify_graph_edges(); ) 1824 1825 Node_List proj_list; 1826 Matcher m(proj_list); 1827 _matcher = &m; 1828 { 1829 TracePhase t2("matcher", &_t_matcher, true); 1830 m.match(); 1831 } 1832 // In debug mode can dump m._nodes.dump() for mapping of ideal to machine 1833 // nodes. Mapping is only valid at the root of each matched subtree. 1834 NOT_PRODUCT( verify_graph_edges(); ) 1835 1836 // If you have too many nodes, or if matching has failed, bail out 1837 check_node_count(0, "out of nodes matching instructions"); 1838 if (failing()) return; 1839 1840 // Build a proper-looking CFG 1841 PhaseCFG cfg(node_arena(), root(), m); 1842 _cfg = &cfg; 1843 { 1844 NOT_PRODUCT( TracePhase t2("scheduler", &_t_scheduler, TimeCompiler); ) 1845 cfg.Dominators(); 1846 if (failing()) return; 1847 1848 NOT_PRODUCT( verify_graph_edges(); ) 1849 1850 cfg.Estimate_Block_Frequency(); 1851 cfg.GlobalCodeMotion(m,unique(),proj_list); 1852 1853 print_method("Global code motion", 2); 1854 1855 if (failing()) return; 1856 NOT_PRODUCT( verify_graph_edges(); ) 1857 1858 debug_only( cfg.verify(); ) 1859 } 1860 NOT_PRODUCT( verify_graph_edges(); ) 1861 1862 PhaseChaitin regalloc(unique(),cfg,m); 1863 _regalloc = ®alloc; 1864 { 1865 TracePhase t2("regalloc", &_t_registerAllocation, true); 1866 // Perform any platform dependent preallocation actions. This is used, 1867 // for example, to avoid taking an implicit null pointer exception 1868 // using the frame pointer on win95. 1869 _regalloc->pd_preallocate_hook(); 1870 1871 // Perform register allocation. After Chaitin, use-def chains are 1872 // no longer accurate (at spill code) and so must be ignored. 1873 // Node->LRG->reg mappings are still accurate. 1874 _regalloc->Register_Allocate(); 1875 1876 // Bail out if the allocator builds too many nodes 1877 if (failing()) return; 1878 } 1879 1880 // Prior to register allocation we kept empty basic blocks in case the 1881 // the allocator needed a place to spill. After register allocation we 1882 // are not adding any new instructions. If any basic block is empty, we 1883 // can now safely remove it. 1884 { 1885 NOT_PRODUCT( TracePhase t2("blockOrdering", &_t_blockOrdering, TimeCompiler); ) 1886 cfg.remove_empty(); 1887 if (do_freq_based_layout()) { 1888 PhaseBlockLayout layout(cfg); 1889 } else { 1890 cfg.set_loop_alignment(); 1891 } 1892 cfg.fixup_flow(); 1893 } 1894 1895 // Perform any platform dependent postallocation verifications. 1896 debug_only( _regalloc->pd_postallocate_verify_hook(); ) 1897 1898 // Apply peephole optimizations 1899 if( OptoPeephole ) { 1900 NOT_PRODUCT( TracePhase t2("peephole", &_t_peephole, TimeCompiler); ) 1901 PhasePeephole peep( _regalloc, cfg); 1902 peep.do_transform(); 1903 } 1904 1905 // Convert Nodes to instruction bits in a buffer 1906 { 1907 // %%%% workspace merge brought two timers together for one job 1908 TracePhase t2a("output", &_t_output, true); 1909 NOT_PRODUCT( TraceTime t2b(NULL, &_t_codeGeneration, TimeCompiler, false); ) 1910 Output(); 1911 } 1912 1913 print_method("Final Code"); 1914 1915 // He's dead, Jim. 1916 _cfg = (PhaseCFG*)0xdeadbeef; 1917 _regalloc = (PhaseChaitin*)0xdeadbeef; 1918 } 1919 1920 1921 //------------------------------dump_asm--------------------------------------- 1922 // Dump formatted assembly 1923 #ifndef PRODUCT 1924 void Compile::dump_asm(int *pcs, uint pc_limit) { 1925 bool cut_short = false; 1926 tty->print_cr("#"); 1927 tty->print("# "); _tf->dump(); tty->cr(); 1928 tty->print_cr("#"); 1929 1930 // For all blocks 1931 int pc = 0x0; // Program counter 1932 char starts_bundle = ' '; 1933 _regalloc->dump_frame(); 1934 1935 Node *n = NULL; 1936 for( uint i=0; i<_cfg->_num_blocks; i++ ) { 1937 if (VMThread::should_terminate()) { cut_short = true; break; } 1938 Block *b = _cfg->_blocks[i]; 1939 if (b->is_connector() && !Verbose) continue; 1940 n = b->_nodes[0]; 1941 if (pcs && n->_idx < pc_limit) 1942 tty->print("%3.3x ", pcs[n->_idx]); 1943 else 1944 tty->print(" "); 1945 b->dump_head( &_cfg->_bbs ); 1946 if (b->is_connector()) { 1947 tty->print_cr(" # Empty connector block"); 1948 } else if (b->num_preds() == 2 && b->pred(1)->is_CatchProj() && b->pred(1)->as_CatchProj()->_con == CatchProjNode::fall_through_index) { 1949 tty->print_cr(" # Block is sole successor of call"); 1950 } 1951 1952 // For all instructions 1953 Node *delay = NULL; 1954 for( uint j = 0; j<b->_nodes.size(); j++ ) { 1955 if (VMThread::should_terminate()) { cut_short = true; break; } 1956 n = b->_nodes[j]; 1957 if (valid_bundle_info(n)) { 1958 Bundle *bundle = node_bundling(n); 1959 if (bundle->used_in_unconditional_delay()) { 1960 delay = n; 1961 continue; 1962 } 1963 if (bundle->starts_bundle()) 1964 starts_bundle = '+'; 1965 } 1966 1967 if (WizardMode) n->dump(); 1968 1969 if( !n->is_Region() && // Dont print in the Assembly 1970 !n->is_Phi() && // a few noisely useless nodes 1971 !n->is_Proj() && 1972 !n->is_MachTemp() && 1973 !n->is_SafePointScalarObject() && 1974 !n->is_Catch() && // Would be nice to print exception table targets 1975 !n->is_MergeMem() && // Not very interesting 1976 !n->is_top() && // Debug info table constants 1977 !(n->is_Con() && !n->is_Mach())// Debug info table constants 1978 ) { 1979 if (pcs && n->_idx < pc_limit) 1980 tty->print("%3.3x", pcs[n->_idx]); 1981 else 1982 tty->print(" "); 1983 tty->print(" %c ", starts_bundle); 1984 starts_bundle = ' '; 1985 tty->print("\t"); 1986 n->format(_regalloc, tty); 1987 tty->cr(); 1988 } 1989 1990 // If we have an instruction with a delay slot, and have seen a delay, 1991 // then back up and print it 1992 if (valid_bundle_info(n) && node_bundling(n)->use_unconditional_delay()) { 1993 assert(delay != NULL, "no unconditional delay instruction"); 1994 if (WizardMode) delay->dump(); 1995 1996 if (node_bundling(delay)->starts_bundle()) 1997 starts_bundle = '+'; 1998 if (pcs && n->_idx < pc_limit) 1999 tty->print("%3.3x", pcs[n->_idx]); 2000 else 2001 tty->print(" "); 2002 tty->print(" %c ", starts_bundle); 2003 starts_bundle = ' '; 2004 tty->print("\t"); 2005 delay->format(_regalloc, tty); 2006 tty->print_cr(""); 2007 delay = NULL; 2008 } 2009 2010 // Dump the exception table as well 2011 if( n->is_Catch() && (Verbose || WizardMode) ) { 2012 // Print the exception table for this offset 2013 _handler_table.print_subtable_for(pc); 2014 } 2015 } 2016 2017 if (pcs && n->_idx < pc_limit) 2018 tty->print_cr("%3.3x", pcs[n->_idx]); 2019 else 2020 tty->print_cr(""); 2021 2022 assert(cut_short || delay == NULL, "no unconditional delay branch"); 2023 2024 } // End of per-block dump 2025 tty->print_cr(""); 2026 2027 if (cut_short) tty->print_cr("*** disassembly is cut short ***"); 2028 } 2029 #endif 2030 2031 //------------------------------Final_Reshape_Counts--------------------------- 2032 // This class defines counters to help identify when a method 2033 // may/must be executed using hardware with only 24-bit precision. 2034 struct Final_Reshape_Counts : public StackObj { 2035 int _call_count; // count non-inlined 'common' calls 2036 int _float_count; // count float ops requiring 24-bit precision 2037 int _double_count; // count double ops requiring more precision 2038 int _java_call_count; // count non-inlined 'java' calls 2039 int _inner_loop_count; // count loops which need alignment 2040 VectorSet _visited; // Visitation flags 2041 Node_List _tests; // Set of IfNodes & PCTableNodes 2042 2043 Final_Reshape_Counts() : 2044 _call_count(0), _float_count(0), _double_count(0), 2045 _java_call_count(0), _inner_loop_count(0), 2046 _visited( Thread::current()->resource_area() ) { } 2047 2048 void inc_call_count () { _call_count ++; } 2049 void inc_float_count () { _float_count ++; } 2050 void inc_double_count() { _double_count++; } 2051 void inc_java_call_count() { _java_call_count++; } 2052 void inc_inner_loop_count() { _inner_loop_count++; } 2053 2054 int get_call_count () const { return _call_count ; } 2055 int get_float_count () const { return _float_count ; } 2056 int get_double_count() const { return _double_count; } 2057 int get_java_call_count() const { return _java_call_count; } 2058 int get_inner_loop_count() const { return _inner_loop_count; } 2059 }; 2060 2061 static bool oop_offset_is_sane(const TypeInstPtr* tp) { 2062 ciInstanceKlass *k = tp->klass()->as_instance_klass(); 2063 // Make sure the offset goes inside the instance layout. 2064 return k->contains_field_offset(tp->offset()); 2065 // Note that OffsetBot and OffsetTop are very negative. 2066 } 2067 2068 //------------------------------final_graph_reshaping_impl---------------------- 2069 // Implement items 1-5 from final_graph_reshaping below. 2070 static void final_graph_reshaping_impl( Node *n, Final_Reshape_Counts &frc ) { 2071 2072 if ( n->outcnt() == 0 ) return; // dead node 2073 uint nop = n->Opcode(); 2074 2075 // Check for 2-input instruction with "last use" on right input. 2076 // Swap to left input. Implements item (2). 2077 if( n->req() == 3 && // two-input instruction 2078 n->in(1)->outcnt() > 1 && // left use is NOT a last use 2079 (!n->in(1)->is_Phi() || n->in(1)->in(2) != n) && // it is not data loop 2080 n->in(2)->outcnt() == 1 &&// right use IS a last use 2081 !n->in(2)->is_Con() ) { // right use is not a constant 2082 // Check for commutative opcode 2083 switch( nop ) { 2084 case Op_AddI: case Op_AddF: case Op_AddD: case Op_AddL: 2085 case Op_MaxI: case Op_MinI: 2086 case Op_MulI: case Op_MulF: case Op_MulD: case Op_MulL: 2087 case Op_AndL: case Op_XorL: case Op_OrL: 2088 case Op_AndI: case Op_XorI: case Op_OrI: { 2089 // Move "last use" input to left by swapping inputs 2090 n->swap_edges(1, 2); 2091 break; 2092 } 2093 default: 2094 break; 2095 } 2096 } 2097 2098 #ifdef ASSERT 2099 if( n->is_Mem() ) { 2100 Compile* C = Compile::current(); 2101 int alias_idx = C->get_alias_index(n->as_Mem()->adr_type()); 2102 assert( n->in(0) != NULL || alias_idx != Compile::AliasIdxRaw || 2103 // oop will be recorded in oop map if load crosses safepoint 2104 n->is_Load() && (n->as_Load()->bottom_type()->isa_oopptr() || 2105 LoadNode::is_immutable_value(n->in(MemNode::Address))), 2106 "raw memory operations should have control edge"); 2107 } 2108 #endif 2109 // Count FPU ops and common calls, implements item (3) 2110 switch( nop ) { 2111 // Count all float operations that may use FPU 2112 case Op_AddF: 2113 case Op_SubF: 2114 case Op_MulF: 2115 case Op_DivF: 2116 case Op_NegF: 2117 case Op_ModF: 2118 case Op_ConvI2F: 2119 case Op_ConF: 2120 case Op_CmpF: 2121 case Op_CmpF3: 2122 // case Op_ConvL2F: // longs are split into 32-bit halves 2123 frc.inc_float_count(); 2124 break; 2125 2126 case Op_ConvF2D: 2127 case Op_ConvD2F: 2128 frc.inc_float_count(); 2129 frc.inc_double_count(); 2130 break; 2131 2132 // Count all double operations that may use FPU 2133 case Op_AddD: 2134 case Op_SubD: 2135 case Op_MulD: 2136 case Op_DivD: 2137 case Op_NegD: 2138 case Op_ModD: 2139 case Op_ConvI2D: 2140 case Op_ConvD2I: 2141 // case Op_ConvL2D: // handled by leaf call 2142 // case Op_ConvD2L: // handled by leaf call 2143 case Op_ConD: 2144 case Op_CmpD: 2145 case Op_CmpD3: 2146 frc.inc_double_count(); 2147 break; 2148 case Op_Opaque1: // Remove Opaque Nodes before matching 2149 case Op_Opaque2: // Remove Opaque Nodes before matching 2150 n->subsume_by(n->in(1)); 2151 break; 2152 case Op_CallStaticJava: 2153 case Op_CallJava: 2154 case Op_CallDynamicJava: 2155 frc.inc_java_call_count(); // Count java call site; 2156 case Op_CallRuntime: 2157 case Op_CallLeaf: 2158 case Op_CallLeafNoFP: { 2159 assert( n->is_Call(), "" ); 2160 CallNode *call = n->as_Call(); 2161 // Count call sites where the FP mode bit would have to be flipped. 2162 // Do not count uncommon runtime calls: 2163 // uncommon_trap, _complete_monitor_locking, _complete_monitor_unlocking, 2164 // _new_Java, _new_typeArray, _new_objArray, _rethrow_Java, ... 2165 if( !call->is_CallStaticJava() || !call->as_CallStaticJava()->_name ) { 2166 frc.inc_call_count(); // Count the call site 2167 } else { // See if uncommon argument is shared 2168 Node *n = call->in(TypeFunc::Parms); 2169 int nop = n->Opcode(); 2170 // Clone shared simple arguments to uncommon calls, item (1). 2171 if( n->outcnt() > 1 && 2172 !n->is_Proj() && 2173 nop != Op_CreateEx && 2174 nop != Op_CheckCastPP && 2175 nop != Op_DecodeN && 2176 !n->is_Mem() ) { 2177 Node *x = n->clone(); 2178 call->set_req( TypeFunc::Parms, x ); 2179 } 2180 } 2181 break; 2182 } 2183 2184 case Op_StoreD: 2185 case Op_LoadD: 2186 case Op_LoadD_unaligned: 2187 frc.inc_double_count(); 2188 goto handle_mem; 2189 case Op_StoreF: 2190 case Op_LoadF: 2191 frc.inc_float_count(); 2192 goto handle_mem; 2193 2194 case Op_StoreB: 2195 case Op_StoreC: 2196 case Op_StoreCM: 2197 case Op_StorePConditional: 2198 case Op_StoreI: 2199 case Op_StoreL: 2200 case Op_StoreIConditional: 2201 case Op_StoreLConditional: 2202 case Op_CompareAndSwapI: 2203 case Op_CompareAndSwapL: 2204 case Op_CompareAndSwapP: 2205 case Op_CompareAndSwapN: 2206 case Op_StoreP: 2207 case Op_StoreN: 2208 case Op_LoadB: 2209 case Op_LoadUB: 2210 case Op_LoadUS: 2211 case Op_LoadI: 2212 case Op_LoadUI2L: 2213 case Op_LoadKlass: 2214 case Op_LoadNKlass: 2215 case Op_LoadL: 2216 case Op_LoadL_unaligned: 2217 case Op_LoadPLocked: 2218 case Op_LoadLLocked: 2219 case Op_LoadP: 2220 case Op_LoadN: 2221 case Op_LoadRange: 2222 case Op_LoadS: { 2223 handle_mem: 2224 #ifdef ASSERT 2225 if( VerifyOptoOopOffsets ) { 2226 assert( n->is_Mem(), "" ); 2227 MemNode *mem = (MemNode*)n; 2228 // Check to see if address types have grounded out somehow. 2229 const TypeInstPtr *tp = mem->in(MemNode::Address)->bottom_type()->isa_instptr(); 2230 assert( !tp || oop_offset_is_sane(tp), "" ); 2231 } 2232 #endif 2233 break; 2234 } 2235 2236 case Op_AddP: { // Assert sane base pointers 2237 Node *addp = n->in(AddPNode::Address); 2238 assert( !addp->is_AddP() || 2239 addp->in(AddPNode::Base)->is_top() || // Top OK for allocation 2240 addp->in(AddPNode::Base) == n->in(AddPNode::Base), 2241 "Base pointers must match" ); 2242 #ifdef _LP64 2243 if (UseCompressedOops && 2244 addp->Opcode() == Op_ConP && 2245 addp == n->in(AddPNode::Base) && 2246 n->in(AddPNode::Offset)->is_Con()) { 2247 // Use addressing with narrow klass to load with offset on x86. 2248 // On sparc loading 32-bits constant and decoding it have less 2249 // instructions (4) then load 64-bits constant (7). 2250 // Do this transformation here since IGVN will convert ConN back to ConP. 2251 const Type* t = addp->bottom_type(); 2252 if (t->isa_oopptr()) { 2253 Node* nn = NULL; 2254 2255 // Look for existing ConN node of the same exact type. 2256 Compile* C = Compile::current(); 2257 Node* r = C->root(); 2258 uint cnt = r->outcnt(); 2259 for (uint i = 0; i < cnt; i++) { 2260 Node* m = r->raw_out(i); 2261 if (m!= NULL && m->Opcode() == Op_ConN && 2262 m->bottom_type()->make_ptr() == t) { 2263 nn = m; 2264 break; 2265 } 2266 } 2267 if (nn != NULL) { 2268 // Decode a narrow oop to match address 2269 // [R12 + narrow_oop_reg<<3 + offset] 2270 nn = new (C, 2) DecodeNNode(nn, t); 2271 n->set_req(AddPNode::Base, nn); 2272 n->set_req(AddPNode::Address, nn); 2273 if (addp->outcnt() == 0) { 2274 addp->disconnect_inputs(NULL); 2275 } 2276 } 2277 } 2278 } 2279 #endif 2280 break; 2281 } 2282 2283 #ifdef _LP64 2284 case Op_CastPP: 2285 if (n->in(1)->is_DecodeN() && Matcher::gen_narrow_oop_implicit_null_checks()) { 2286 Compile* C = Compile::current(); 2287 Node* in1 = n->in(1); 2288 const Type* t = n->bottom_type(); 2289 Node* new_in1 = in1->clone(); 2290 new_in1->as_DecodeN()->set_type(t); 2291 2292 if (!Matcher::narrow_oop_use_complex_address()) { 2293 // 2294 // x86, ARM and friends can handle 2 adds in addressing mode 2295 // and Matcher can fold a DecodeN node into address by using 2296 // a narrow oop directly and do implicit NULL check in address: 2297 // 2298 // [R12 + narrow_oop_reg<<3 + offset] 2299 // NullCheck narrow_oop_reg 2300 // 2301 // On other platforms (Sparc) we have to keep new DecodeN node and 2302 // use it to do implicit NULL check in address: 2303 // 2304 // decode_not_null narrow_oop_reg, base_reg 2305 // [base_reg + offset] 2306 // NullCheck base_reg 2307 // 2308 // Pin the new DecodeN node to non-null path on these platform (Sparc) 2309 // to keep the information to which NULL check the new DecodeN node 2310 // corresponds to use it as value in implicit_null_check(). 2311 // 2312 new_in1->set_req(0, n->in(0)); 2313 } 2314 2315 n->subsume_by(new_in1); 2316 if (in1->outcnt() == 0) { 2317 in1->disconnect_inputs(NULL); 2318 } 2319 } 2320 break; 2321 2322 case Op_CmpP: 2323 // Do this transformation here to preserve CmpPNode::sub() and 2324 // other TypePtr related Ideal optimizations (for example, ptr nullness). 2325 if (n->in(1)->is_DecodeN() || n->in(2)->is_DecodeN()) { 2326 Node* in1 = n->in(1); 2327 Node* in2 = n->in(2); 2328 if (!in1->is_DecodeN()) { 2329 in2 = in1; 2330 in1 = n->in(2); 2331 } 2332 assert(in1->is_DecodeN(), "sanity"); 2333 2334 Compile* C = Compile::current(); 2335 Node* new_in2 = NULL; 2336 if (in2->is_DecodeN()) { 2337 new_in2 = in2->in(1); 2338 } else if (in2->Opcode() == Op_ConP) { 2339 const Type* t = in2->bottom_type(); 2340 if (t == TypePtr::NULL_PTR) { 2341 // Don't convert CmpP null check into CmpN if compressed 2342 // oops implicit null check is not generated. 2343 // This will allow to generate normal oop implicit null check. 2344 if (Matcher::gen_narrow_oop_implicit_null_checks()) 2345 new_in2 = ConNode::make(C, TypeNarrowOop::NULL_PTR); 2346 // 2347 // This transformation together with CastPP transformation above 2348 // will generated code for implicit NULL checks for compressed oops. 2349 // 2350 // The original code after Optimize() 2351 // 2352 // LoadN memory, narrow_oop_reg 2353 // decode narrow_oop_reg, base_reg 2354 // CmpP base_reg, NULL 2355 // CastPP base_reg // NotNull 2356 // Load [base_reg + offset], val_reg 2357 // 2358 // after these transformations will be 2359 // 2360 // LoadN memory, narrow_oop_reg 2361 // CmpN narrow_oop_reg, NULL 2362 // decode_not_null narrow_oop_reg, base_reg 2363 // Load [base_reg + offset], val_reg 2364 // 2365 // and the uncommon path (== NULL) will use narrow_oop_reg directly 2366 // since narrow oops can be used in debug info now (see the code in 2367 // final_graph_reshaping_walk()). 2368 // 2369 // At the end the code will be matched to 2370 // on x86: 2371 // 2372 // Load_narrow_oop memory, narrow_oop_reg 2373 // Load [R12 + narrow_oop_reg<<3 + offset], val_reg 2374 // NullCheck narrow_oop_reg 2375 // 2376 // and on sparc: 2377 // 2378 // Load_narrow_oop memory, narrow_oop_reg 2379 // decode_not_null narrow_oop_reg, base_reg 2380 // Load [base_reg + offset], val_reg 2381 // NullCheck base_reg 2382 // 2383 } else if (t->isa_oopptr()) { 2384 new_in2 = ConNode::make(C, t->make_narrowoop()); 2385 } 2386 } 2387 if (new_in2 != NULL) { 2388 Node* cmpN = new (C, 3) CmpNNode(in1->in(1), new_in2); 2389 n->subsume_by( cmpN ); 2390 if (in1->outcnt() == 0) { 2391 in1->disconnect_inputs(NULL); 2392 } 2393 if (in2->outcnt() == 0) { 2394 in2->disconnect_inputs(NULL); 2395 } 2396 } 2397 } 2398 break; 2399 2400 case Op_DecodeN: 2401 assert(!n->in(1)->is_EncodeP(), "should be optimized out"); 2402 // DecodeN could be pinned when it can't be fold into 2403 // an address expression, see the code for Op_CastPP above. 2404 assert(n->in(0) == NULL || !Matcher::narrow_oop_use_complex_address(), "no control"); 2405 break; 2406 2407 case Op_EncodeP: { 2408 Node* in1 = n->in(1); 2409 if (in1->is_DecodeN()) { 2410 n->subsume_by(in1->in(1)); 2411 } else if (in1->Opcode() == Op_ConP) { 2412 Compile* C = Compile::current(); 2413 const Type* t = in1->bottom_type(); 2414 if (t == TypePtr::NULL_PTR) { 2415 n->subsume_by(ConNode::make(C, TypeNarrowOop::NULL_PTR)); 2416 } else if (t->isa_oopptr()) { 2417 n->subsume_by(ConNode::make(C, t->make_narrowoop())); 2418 } 2419 } 2420 if (in1->outcnt() == 0) { 2421 in1->disconnect_inputs(NULL); 2422 } 2423 break; 2424 } 2425 2426 case Op_Proj: { 2427 if (OptimizeStringConcat) { 2428 ProjNode* p = n->as_Proj(); 2429 if (p->_is_io_use) { 2430 // Separate projections were used for the exception path which 2431 // are normally removed by a late inline. If it wasn't inlined 2432 // then they will hang around and should just be replaced with 2433 // the original one. 2434 Node* proj = NULL; 2435 // Replace with just one 2436 for (SimpleDUIterator i(p->in(0)); i.has_next(); i.next()) { 2437 Node *use = i.get(); 2438 if (use->is_Proj() && p != use && use->as_Proj()->_con == p->_con) { 2439 proj = use; 2440 break; 2441 } 2442 } 2443 assert(p != NULL, "must be found"); 2444 p->subsume_by(proj); 2445 } 2446 } 2447 break; 2448 } 2449 2450 case Op_Phi: 2451 if (n->as_Phi()->bottom_type()->isa_narrowoop()) { 2452 // The EncodeP optimization may create Phi with the same edges 2453 // for all paths. It is not handled well by Register Allocator. 2454 Node* unique_in = n->in(1); 2455 assert(unique_in != NULL, ""); 2456 uint cnt = n->req(); 2457 for (uint i = 2; i < cnt; i++) { 2458 Node* m = n->in(i); 2459 assert(m != NULL, ""); 2460 if (unique_in != m) 2461 unique_in = NULL; 2462 } 2463 if (unique_in != NULL) { 2464 n->subsume_by(unique_in); 2465 } 2466 } 2467 break; 2468 2469 #endif 2470 2471 case Op_ModI: 2472 if (UseDivMod) { 2473 // Check if a%b and a/b both exist 2474 Node* d = n->find_similar(Op_DivI); 2475 if (d) { 2476 // Replace them with a fused divmod if supported 2477 Compile* C = Compile::current(); 2478 if (Matcher::has_match_rule(Op_DivModI)) { 2479 DivModINode* divmod = DivModINode::make(C, n); 2480 d->subsume_by(divmod->div_proj()); 2481 n->subsume_by(divmod->mod_proj()); 2482 } else { 2483 // replace a%b with a-((a/b)*b) 2484 Node* mult = new (C, 3) MulINode(d, d->in(2)); 2485 Node* sub = new (C, 3) SubINode(d->in(1), mult); 2486 n->subsume_by( sub ); 2487 } 2488 } 2489 } 2490 break; 2491 2492 case Op_ModL: 2493 if (UseDivMod) { 2494 // Check if a%b and a/b both exist 2495 Node* d = n->find_similar(Op_DivL); 2496 if (d) { 2497 // Replace them with a fused divmod if supported 2498 Compile* C = Compile::current(); 2499 if (Matcher::has_match_rule(Op_DivModL)) { 2500 DivModLNode* divmod = DivModLNode::make(C, n); 2501 d->subsume_by(divmod->div_proj()); 2502 n->subsume_by(divmod->mod_proj()); 2503 } else { 2504 // replace a%b with a-((a/b)*b) 2505 Node* mult = new (C, 3) MulLNode(d, d->in(2)); 2506 Node* sub = new (C, 3) SubLNode(d->in(1), mult); 2507 n->subsume_by( sub ); 2508 } 2509 } 2510 } 2511 break; 2512 2513 case Op_Load16B: 2514 case Op_Load8B: 2515 case Op_Load4B: 2516 case Op_Load8S: 2517 case Op_Load4S: 2518 case Op_Load2S: 2519 case Op_Load8C: 2520 case Op_Load4C: 2521 case Op_Load2C: 2522 case Op_Load4I: 2523 case Op_Load2I: 2524 case Op_Load2L: 2525 case Op_Load4F: 2526 case Op_Load2F: 2527 case Op_Load2D: 2528 case Op_Store16B: 2529 case Op_Store8B: 2530 case Op_Store4B: 2531 case Op_Store8C: 2532 case Op_Store4C: 2533 case Op_Store2C: 2534 case Op_Store4I: 2535 case Op_Store2I: 2536 case Op_Store2L: 2537 case Op_Store4F: 2538 case Op_Store2F: 2539 case Op_Store2D: 2540 break; 2541 2542 case Op_PackB: 2543 case Op_PackS: 2544 case Op_PackC: 2545 case Op_PackI: 2546 case Op_PackF: 2547 case Op_PackL: 2548 case Op_PackD: 2549 if (n->req()-1 > 2) { 2550 // Replace many operand PackNodes with a binary tree for matching 2551 PackNode* p = (PackNode*) n; 2552 Node* btp = p->binaryTreePack(Compile::current(), 1, n->req()); 2553 n->subsume_by(btp); 2554 } 2555 break; 2556 case Op_Loop: 2557 case Op_CountedLoop: 2558 if (n->as_Loop()->is_inner_loop()) { 2559 frc.inc_inner_loop_count(); 2560 } 2561 break; 2562 default: 2563 assert( !n->is_Call(), "" ); 2564 assert( !n->is_Mem(), "" ); 2565 break; 2566 } 2567 2568 // Collect CFG split points 2569 if (n->is_MultiBranch()) 2570 frc._tests.push(n); 2571 } 2572 2573 //------------------------------final_graph_reshaping_walk--------------------- 2574 // Replacing Opaque nodes with their input in final_graph_reshaping_impl(), 2575 // requires that the walk visits a node's inputs before visiting the node. 2576 static void final_graph_reshaping_walk( Node_Stack &nstack, Node *root, Final_Reshape_Counts &frc ) { 2577 ResourceArea *area = Thread::current()->resource_area(); 2578 Unique_Node_List sfpt(area); 2579 2580 frc._visited.set(root->_idx); // first, mark node as visited 2581 uint cnt = root->req(); 2582 Node *n = root; 2583 uint i = 0; 2584 while (true) { 2585 if (i < cnt) { 2586 // Place all non-visited non-null inputs onto stack 2587 Node* m = n->in(i); 2588 ++i; 2589 if (m != NULL && !frc._visited.test_set(m->_idx)) { 2590 if (m->is_SafePoint() && m->as_SafePoint()->jvms() != NULL) 2591 sfpt.push(m); 2592 cnt = m->req(); 2593 nstack.push(n, i); // put on stack parent and next input's index 2594 n = m; 2595 i = 0; 2596 } 2597 } else { 2598 // Now do post-visit work 2599 final_graph_reshaping_impl( n, frc ); 2600 if (nstack.is_empty()) 2601 break; // finished 2602 n = nstack.node(); // Get node from stack 2603 cnt = n->req(); 2604 i = nstack.index(); 2605 nstack.pop(); // Shift to the next node on stack 2606 } 2607 } 2608 2609 // Skip next transformation if compressed oops are not used. 2610 if (!UseCompressedOops || !Matcher::gen_narrow_oop_implicit_null_checks()) 2611 return; 2612 2613 // Go over safepoints nodes to skip DecodeN nodes for debug edges. 2614 // It could be done for an uncommon traps or any safepoints/calls 2615 // if the DecodeN node is referenced only in a debug info. 2616 while (sfpt.size() > 0) { 2617 n = sfpt.pop(); 2618 JVMState *jvms = n->as_SafePoint()->jvms(); 2619 assert(jvms != NULL, "sanity"); 2620 int start = jvms->debug_start(); 2621 int end = n->req(); 2622 bool is_uncommon = (n->is_CallStaticJava() && 2623 n->as_CallStaticJava()->uncommon_trap_request() != 0); 2624 for (int j = start; j < end; j++) { 2625 Node* in = n->in(j); 2626 if (in->is_DecodeN()) { 2627 bool safe_to_skip = true; 2628 if (!is_uncommon ) { 2629 // Is it safe to skip? 2630 for (uint i = 0; i < in->outcnt(); i++) { 2631 Node* u = in->raw_out(i); 2632 if (!u->is_SafePoint() || 2633 u->is_Call() && u->as_Call()->has_non_debug_use(n)) { 2634 safe_to_skip = false; 2635 } 2636 } 2637 } 2638 if (safe_to_skip) { 2639 n->set_req(j, in->in(1)); 2640 } 2641 if (in->outcnt() == 0) { 2642 in->disconnect_inputs(NULL); 2643 } 2644 } 2645 } 2646 } 2647 } 2648 2649 //------------------------------final_graph_reshaping-------------------------- 2650 // Final Graph Reshaping. 2651 // 2652 // (1) Clone simple inputs to uncommon calls, so they can be scheduled late 2653 // and not commoned up and forced early. Must come after regular 2654 // optimizations to avoid GVN undoing the cloning. Clone constant 2655 // inputs to Loop Phis; these will be split by the allocator anyways. 2656 // Remove Opaque nodes. 2657 // (2) Move last-uses by commutative operations to the left input to encourage 2658 // Intel update-in-place two-address operations and better register usage 2659 // on RISCs. Must come after regular optimizations to avoid GVN Ideal 2660 // calls canonicalizing them back. 2661 // (3) Count the number of double-precision FP ops, single-precision FP ops 2662 // and call sites. On Intel, we can get correct rounding either by 2663 // forcing singles to memory (requires extra stores and loads after each 2664 // FP bytecode) or we can set a rounding mode bit (requires setting and 2665 // clearing the mode bit around call sites). The mode bit is only used 2666 // if the relative frequency of single FP ops to calls is low enough. 2667 // This is a key transform for SPEC mpeg_audio. 2668 // (4) Detect infinite loops; blobs of code reachable from above but not 2669 // below. Several of the Code_Gen algorithms fail on such code shapes, 2670 // so we simply bail out. Happens a lot in ZKM.jar, but also happens 2671 // from time to time in other codes (such as -Xcomp finalizer loops, etc). 2672 // Detection is by looking for IfNodes where only 1 projection is 2673 // reachable from below or CatchNodes missing some targets. 2674 // (5) Assert for insane oop offsets in debug mode. 2675 2676 bool Compile::final_graph_reshaping() { 2677 // an infinite loop may have been eliminated by the optimizer, 2678 // in which case the graph will be empty. 2679 if (root()->req() == 1) { 2680 record_method_not_compilable("trivial infinite loop"); 2681 return true; 2682 } 2683 2684 Final_Reshape_Counts frc; 2685 2686 // Visit everybody reachable! 2687 // Allocate stack of size C->unique()/2 to avoid frequent realloc 2688 Node_Stack nstack(unique() >> 1); 2689 final_graph_reshaping_walk(nstack, root(), frc); 2690 2691 // Check for unreachable (from below) code (i.e., infinite loops). 2692 for( uint i = 0; i < frc._tests.size(); i++ ) { 2693 MultiBranchNode *n = frc._tests[i]->as_MultiBranch(); 2694 // Get number of CFG targets. 2695 // Note that PCTables include exception targets after calls. 2696 uint required_outcnt = n->required_outcnt(); 2697 if (n->outcnt() != required_outcnt) { 2698 // Check for a few special cases. Rethrow Nodes never take the 2699 // 'fall-thru' path, so expected kids is 1 less. 2700 if (n->is_PCTable() && n->in(0) && n->in(0)->in(0)) { 2701 if (n->in(0)->in(0)->is_Call()) { 2702 CallNode *call = n->in(0)->in(0)->as_Call(); 2703 if (call->entry_point() == OptoRuntime::rethrow_stub()) { 2704 required_outcnt--; // Rethrow always has 1 less kid 2705 } else if (call->req() > TypeFunc::Parms && 2706 call->is_CallDynamicJava()) { 2707 // Check for null receiver. In such case, the optimizer has 2708 // detected that the virtual call will always result in a null 2709 // pointer exception. The fall-through projection of this CatchNode 2710 // will not be populated. 2711 Node *arg0 = call->in(TypeFunc::Parms); 2712 if (arg0->is_Type() && 2713 arg0->as_Type()->type()->higher_equal(TypePtr::NULL_PTR)) { 2714 required_outcnt--; 2715 } 2716 } else if (call->entry_point() == OptoRuntime::new_array_Java() && 2717 call->req() > TypeFunc::Parms+1 && 2718 call->is_CallStaticJava()) { 2719 // Check for negative array length. In such case, the optimizer has 2720 // detected that the allocation attempt will always result in an 2721 // exception. There is no fall-through projection of this CatchNode . 2722 Node *arg1 = call->in(TypeFunc::Parms+1); 2723 if (arg1->is_Type() && 2724 arg1->as_Type()->type()->join(TypeInt::POS)->empty()) { 2725 required_outcnt--; 2726 } 2727 } 2728 } 2729 } 2730 // Recheck with a better notion of 'required_outcnt' 2731 if (n->outcnt() != required_outcnt) { 2732 record_method_not_compilable("malformed control flow"); 2733 return true; // Not all targets reachable! 2734 } 2735 } 2736 // Check that I actually visited all kids. Unreached kids 2737 // must be infinite loops. 2738 for (DUIterator_Fast jmax, j = n->fast_outs(jmax); j < jmax; j++) 2739 if (!frc._visited.test(n->fast_out(j)->_idx)) { 2740 record_method_not_compilable("infinite loop"); 2741 return true; // Found unvisited kid; must be unreach 2742 } 2743 } 2744 2745 // If original bytecodes contained a mixture of floats and doubles 2746 // check if the optimizer has made it homogenous, item (3). 2747 if( Use24BitFPMode && Use24BitFP && UseSSE == 0 && 2748 frc.get_float_count() > 32 && 2749 frc.get_double_count() == 0 && 2750 (10 * frc.get_call_count() < frc.get_float_count()) ) { 2751 set_24_bit_selection_and_mode( false, true ); 2752 } 2753 2754 set_java_calls(frc.get_java_call_count()); 2755 set_inner_loops(frc.get_inner_loop_count()); 2756 2757 // No infinite loops, no reason to bail out. 2758 return false; 2759 } 2760 2761 //-----------------------------too_many_traps---------------------------------- 2762 // Report if there are too many traps at the current method and bci. 2763 // Return true if there was a trap, and/or PerMethodTrapLimit is exceeded. 2764 bool Compile::too_many_traps(ciMethod* method, 2765 int bci, 2766 Deoptimization::DeoptReason reason) { 2767 ciMethodData* md = method->method_data(); 2768 if (md->is_empty()) { 2769 // Assume the trap has not occurred, or that it occurred only 2770 // because of a transient condition during start-up in the interpreter. 2771 return false; 2772 } 2773 if (md->has_trap_at(bci, reason) != 0) { 2774 // Assume PerBytecodeTrapLimit==0, for a more conservative heuristic. 2775 // Also, if there are multiple reasons, or if there is no per-BCI record, 2776 // assume the worst. 2777 if (log()) 2778 log()->elem("observe trap='%s' count='%d'", 2779 Deoptimization::trap_reason_name(reason), 2780 md->trap_count(reason)); 2781 return true; 2782 } else { 2783 // Ignore method/bci and see if there have been too many globally. 2784 return too_many_traps(reason, md); 2785 } 2786 } 2787 2788 // Less-accurate variant which does not require a method and bci. 2789 bool Compile::too_many_traps(Deoptimization::DeoptReason reason, 2790 ciMethodData* logmd) { 2791 if (trap_count(reason) >= (uint)PerMethodTrapLimit) { 2792 // Too many traps globally. 2793 // Note that we use cumulative trap_count, not just md->trap_count. 2794 if (log()) { 2795 int mcount = (logmd == NULL)? -1: (int)logmd->trap_count(reason); 2796 log()->elem("observe trap='%s' count='0' mcount='%d' ccount='%d'", 2797 Deoptimization::trap_reason_name(reason), 2798 mcount, trap_count(reason)); 2799 } 2800 return true; 2801 } else { 2802 // The coast is clear. 2803 return false; 2804 } 2805 } 2806 2807 //--------------------------too_many_recompiles-------------------------------- 2808 // Report if there are too many recompiles at the current method and bci. 2809 // Consults PerBytecodeRecompilationCutoff and PerMethodRecompilationCutoff. 2810 // Is not eager to return true, since this will cause the compiler to use 2811 // Action_none for a trap point, to avoid too many recompilations. 2812 bool Compile::too_many_recompiles(ciMethod* method, 2813 int bci, 2814 Deoptimization::DeoptReason reason) { 2815 ciMethodData* md = method->method_data(); 2816 if (md->is_empty()) { 2817 // Assume the trap has not occurred, or that it occurred only 2818 // because of a transient condition during start-up in the interpreter. 2819 return false; 2820 } 2821 // Pick a cutoff point well within PerBytecodeRecompilationCutoff. 2822 uint bc_cutoff = (uint) PerBytecodeRecompilationCutoff / 8; 2823 uint m_cutoff = (uint) PerMethodRecompilationCutoff / 2 + 1; // not zero 2824 Deoptimization::DeoptReason per_bc_reason 2825 = Deoptimization::reason_recorded_per_bytecode_if_any(reason); 2826 if ((per_bc_reason == Deoptimization::Reason_none 2827 || md->has_trap_at(bci, reason) != 0) 2828 // The trap frequency measure we care about is the recompile count: 2829 && md->trap_recompiled_at(bci) 2830 && md->overflow_recompile_count() >= bc_cutoff) { 2831 // Do not emit a trap here if it has already caused recompilations. 2832 // Also, if there are multiple reasons, or if there is no per-BCI record, 2833 // assume the worst. 2834 if (log()) 2835 log()->elem("observe trap='%s recompiled' count='%d' recompiles2='%d'", 2836 Deoptimization::trap_reason_name(reason), 2837 md->trap_count(reason), 2838 md->overflow_recompile_count()); 2839 return true; 2840 } else if (trap_count(reason) != 0 2841 && decompile_count() >= m_cutoff) { 2842 // Too many recompiles globally, and we have seen this sort of trap. 2843 // Use cumulative decompile_count, not just md->decompile_count. 2844 if (log()) 2845 log()->elem("observe trap='%s' count='%d' mcount='%d' decompiles='%d' mdecompiles='%d'", 2846 Deoptimization::trap_reason_name(reason), 2847 md->trap_count(reason), trap_count(reason), 2848 md->decompile_count(), decompile_count()); 2849 return true; 2850 } else { 2851 // The coast is clear. 2852 return false; 2853 } 2854 } 2855 2856 2857 #ifndef PRODUCT 2858 //------------------------------verify_graph_edges--------------------------- 2859 // Walk the Graph and verify that there is a one-to-one correspondence 2860 // between Use-Def edges and Def-Use edges in the graph. 2861 void Compile::verify_graph_edges(bool no_dead_code) { 2862 if (VerifyGraphEdges) { 2863 ResourceArea *area = Thread::current()->resource_area(); 2864 Unique_Node_List visited(area); 2865 // Call recursive graph walk to check edges 2866 _root->verify_edges(visited); 2867 if (no_dead_code) { 2868 // Now make sure that no visited node is used by an unvisited node. 2869 bool dead_nodes = 0; 2870 Unique_Node_List checked(area); 2871 while (visited.size() > 0) { 2872 Node* n = visited.pop(); 2873 checked.push(n); 2874 for (uint i = 0; i < n->outcnt(); i++) { 2875 Node* use = n->raw_out(i); 2876 if (checked.member(use)) continue; // already checked 2877 if (visited.member(use)) continue; // already in the graph 2878 if (use->is_Con()) continue; // a dead ConNode is OK 2879 // At this point, we have found a dead node which is DU-reachable. 2880 if (dead_nodes++ == 0) 2881 tty->print_cr("*** Dead nodes reachable via DU edges:"); 2882 use->dump(2); 2883 tty->print_cr("---"); 2884 checked.push(use); // No repeats; pretend it is now checked. 2885 } 2886 } 2887 assert(dead_nodes == 0, "using nodes must be reachable from root"); 2888 } 2889 } 2890 } 2891 #endif 2892 2893 // The Compile object keeps track of failure reasons separately from the ciEnv. 2894 // This is required because there is not quite a 1-1 relation between the 2895 // ciEnv and its compilation task and the Compile object. Note that one 2896 // ciEnv might use two Compile objects, if C2Compiler::compile_method decides 2897 // to backtrack and retry without subsuming loads. Other than this backtracking 2898 // behavior, the Compile's failure reason is quietly copied up to the ciEnv 2899 // by the logic in C2Compiler. 2900 void Compile::record_failure(const char* reason) { 2901 if (log() != NULL) { 2902 log()->elem("failure reason='%s' phase='compile'", reason); 2903 } 2904 if (_failure_reason == NULL) { 2905 // Record the first failure reason. 2906 _failure_reason = reason; 2907 } 2908 if (!C->failure_reason_is(C2Compiler::retry_no_subsuming_loads())) { 2909 C->print_method(_failure_reason); 2910 } 2911 _root = NULL; // flush the graph, too 2912 } 2913 2914 Compile::TracePhase::TracePhase(const char* name, elapsedTimer* accumulator, bool dolog) 2915 : TraceTime(NULL, accumulator, false NOT_PRODUCT( || TimeCompiler ), false) 2916 { 2917 if (dolog) { 2918 C = Compile::current(); 2919 _log = C->log(); 2920 } else { 2921 C = NULL; 2922 _log = NULL; 2923 } 2924 if (_log != NULL) { 2925 _log->begin_head("phase name='%s' nodes='%d'", name, C->unique()); 2926 _log->stamp(); 2927 _log->end_head(); 2928 } 2929 } 2930 2931 Compile::TracePhase::~TracePhase() { 2932 if (_log != NULL) { 2933 _log->done("phase nodes='%d'", C->unique()); 2934 } 2935 } 2936 2937 //============================================================================= 2938 // Two Constant's are equal when the type and the value are equal. 2939 bool Compile::Constant::operator==(const Constant& other) { 2940 if (type() != other.type() ) return false; 2941 if (can_be_reused() != other.can_be_reused()) return false; 2942 // For floating point values we compare the bit pattern. 2943 switch (type()) { 2944 case T_FLOAT: return (_value.i == other._value.i); 2945 case T_LONG: 2946 case T_DOUBLE: return (_value.j == other._value.j); 2947 case T_OBJECT: 2948 case T_ADDRESS: return (_value.l == other._value.l); 2949 case T_VOID: return (_value.l == other._value.l); // jump-table entries 2950 default: ShouldNotReachHere(); 2951 } 2952 return false; 2953 } 2954 2955 // Emit constants grouped in the following order: 2956 static BasicType type_order[] = { 2957 T_FLOAT, // 32-bit 2958 T_OBJECT, // 32 or 64-bit 2959 T_ADDRESS, // 32 or 64-bit 2960 T_DOUBLE, // 64-bit 2961 T_LONG, // 64-bit 2962 T_VOID, // 32 or 64-bit (jump-tables are at the end of the constant table for code emission reasons) 2963 T_ILLEGAL 2964 }; 2965 2966 static int type_to_size_in_bytes(BasicType t) { 2967 switch (t) { 2968 case T_LONG: return sizeof(jlong ); 2969 case T_FLOAT: return sizeof(jfloat ); 2970 case T_DOUBLE: return sizeof(jdouble); 2971 // We use T_VOID as marker for jump-table entries (labels) which 2972 // need an interal word relocation. 2973 case T_VOID: 2974 case T_ADDRESS: 2975 case T_OBJECT: return sizeof(jobject); 2976 } 2977 2978 ShouldNotReachHere(); 2979 return -1; 2980 } 2981 2982 void Compile::ConstantTable::calculate_offsets_and_size() { 2983 int size = 0; 2984 for (int t = 0; type_order[t] != T_ILLEGAL; t++) { 2985 BasicType type = type_order[t]; 2986 2987 for (int i = 0; i < _constants.length(); i++) { 2988 Constant con = _constants.at(i); 2989 if (con.type() != type) continue; // Skip other types. 2990 2991 // Align size for type. 2992 int typesize = type_to_size_in_bytes(con.type()); 2993 size = align_size_up(size, typesize); 2994 2995 // Set offset. 2996 con.set_offset(size); 2997 _constants.at_put(i, con); 2998 2999 // Add type size. 3000 size = size + typesize; 3001 } 3002 } 3003 3004 // Align size up to the next section start (which is insts; see 3005 // CodeBuffer::align_at_start). 3006 assert(_size == -1, "already set?"); 3007 _size = align_size_up(size, CodeEntryAlignment); 3008 3009 if (Matcher::constant_table_absolute_addressing) { 3010 set_table_base_offset(0); // No table base offset required 3011 } else { 3012 if (UseRDPCForConstantTableBase) { 3013 // table base offset is set in MachConstantBaseNode::emit 3014 } else { 3015 // When RDPC is not used, the table base is set into the middle of 3016 // the constant table. 3017 int half_size = _size / 2; 3018 assert(half_size * 2 == _size, "sanity"); 3019 set_table_base_offset(-half_size); 3020 } 3021 } 3022 } 3023 3024 void Compile::ConstantTable::emit(CodeBuffer& cb) { 3025 MacroAssembler _masm(&cb); 3026 for (int t = 0; type_order[t] != T_ILLEGAL; t++) { 3027 BasicType type = type_order[t]; 3028 3029 for (int i = 0; i < _constants.length(); i++) { 3030 Constant con = _constants.at(i); 3031 if (con.type() != type) continue; // Skip other types. 3032 3033 address constant_addr; 3034 switch (con.type()) { 3035 case T_LONG: constant_addr = _masm.long_constant( con.get_jlong() ); break; 3036 case T_FLOAT: constant_addr = _masm.float_constant( con.get_jfloat() ); break; 3037 case T_DOUBLE: constant_addr = _masm.double_constant(con.get_jdouble()); break; 3038 case T_OBJECT: { 3039 jobject obj = con.get_jobject(); 3040 int oop_index = _masm.oop_recorder()->find_index(obj); 3041 constant_addr = _masm.address_constant((address) obj, oop_Relocation::spec(oop_index)); 3042 break; 3043 } 3044 case T_ADDRESS: { 3045 address addr = (address) con.get_jobject(); 3046 constant_addr = _masm.address_constant(addr); 3047 break; 3048 } 3049 // We use T_VOID as marker for jump-table entries (labels) which 3050 // need an interal word relocation. 3051 case T_VOID: { 3052 // Write a dummy word. The real value is filled in later 3053 // in fill_jump_table_in_constant_table. 3054 address addr = (address) con.get_jobject(); 3055 constant_addr = _masm.address_constant(addr); 3056 break; 3057 } 3058 default: ShouldNotReachHere(); 3059 } 3060 assert(constant_addr != NULL, "consts section too small"); 3061 assert((constant_addr - _masm.code()->consts()->start()) == con.offset(), err_msg("must be: %d == %d", constant_addr - _masm.code()->consts()->start(), con.offset())); 3062 } 3063 } 3064 } 3065 3066 int Compile::ConstantTable::find_offset(Constant& con) const { 3067 int idx = _constants.find(con); 3068 assert(idx != -1, "constant must be in constant table"); 3069 int offset = _constants.at(idx).offset(); 3070 assert(offset != -1, "constant table not emitted yet?"); 3071 return offset; 3072 } 3073 3074 void Compile::ConstantTable::add(Constant& con) { 3075 if (con.can_be_reused()) { 3076 int idx = _constants.find(con); 3077 if (idx != -1 && _constants.at(idx).can_be_reused()) { 3078 return; 3079 } 3080 } 3081 (void) _constants.append(con); 3082 } 3083 3084 Compile::Constant Compile::ConstantTable::add(BasicType type, jvalue value) { 3085 Constant con(type, value); 3086 add(con); 3087 return con; 3088 } 3089 3090 Compile::Constant Compile::ConstantTable::add(MachOper* oper) { 3091 jvalue value; 3092 BasicType type = oper->type()->basic_type(); 3093 switch (type) { 3094 case T_LONG: value.j = oper->constantL(); break; 3095 case T_FLOAT: value.f = oper->constantF(); break; 3096 case T_DOUBLE: value.d = oper->constantD(); break; 3097 case T_OBJECT: 3098 case T_ADDRESS: value.l = (jobject) oper->constant(); break; 3099 default: ShouldNotReachHere(); 3100 } 3101 return add(type, value); 3102 } 3103 3104 Compile::Constant Compile::ConstantTable::allocate_jump_table(MachConstantNode* n) { 3105 jvalue value; 3106 // We can use the node pointer here to identify the right jump-table 3107 // as this method is called from Compile::Fill_buffer right before 3108 // the MachNodes are emitted and the jump-table is filled (means the 3109 // MachNode pointers do not change anymore). 3110 value.l = (jobject) n; 3111 Constant con(T_VOID, value, false); // Labels of a jump-table cannot be reused. 3112 for (uint i = 0; i < n->outcnt(); i++) { 3113 add(con); 3114 } 3115 return con; 3116 } 3117 3118 void Compile::ConstantTable::fill_jump_table(CodeBuffer& cb, MachConstantNode* n, GrowableArray<Label*> labels) const { 3119 // If called from Compile::scratch_emit_size do nothing. 3120 if (Compile::current()->in_scratch_emit_size()) return; 3121 3122 assert(labels.is_nonempty(), "must be"); 3123 assert((uint) labels.length() == n->outcnt(), err_msg("must be equal: %d == %d", labels.length(), n->outcnt())); 3124 3125 // Since MachConstantNode::constant_offset() also contains 3126 // table_base_offset() we need to subtract the table_base_offset() 3127 // to get the plain offset into the constant table. 3128 int offset = n->constant_offset() - table_base_offset(); 3129 3130 MacroAssembler _masm(&cb); 3131 address* jump_table_base = (address*) (_masm.code()->consts()->start() + offset); 3132 3133 for (int i = 0; i < labels.length(); i++) { 3134 address* constant_addr = &jump_table_base[i]; 3135 assert(*constant_addr == (address) n, "all jump-table entries must contain node pointer"); 3136 *constant_addr = cb.consts()->target(*labels.at(i), (address) constant_addr); 3137 cb.consts()->relocate((address) constant_addr, relocInfo::internal_word_type); 3138 } 3139 }