Mercurial > hg > truffle
annotate src/share/vm/c1/c1_IR.cpp @ 1579:e9ff18c4ace7
Merge
| author | jrose |
|---|---|
| date | Wed, 02 Jun 2010 22:45:42 -0700 |
| parents | c18cbe5936b8 61b2245abf36 |
| children | b812ff5abc73 |
| rev | line source |
|---|---|
| 0 | 1 /* |
|
1552
c18cbe5936b8
6941466: Oracle rebranding changes for Hotspot repositories
trims
parents:
1295
diff
changeset
|
2 * Copyright (c) 1999, 2010, Oracle and/or its affiliates. All rights reserved. |
| 0 | 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 * | |
|
1552
c18cbe5936b8
6941466: Oracle rebranding changes for Hotspot repositories
trims
parents:
1295
diff
changeset
|
19 * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA |
|
c18cbe5936b8
6941466: Oracle rebranding changes for Hotspot repositories
trims
parents:
1295
diff
changeset
|
20 * or visit www.oracle.com if you need additional information or have any |
|
c18cbe5936b8
6941466: Oracle rebranding changes for Hotspot repositories
trims
parents:
1295
diff
changeset
|
21 * questions. |
| 0 | 22 * |
| 23 */ | |
| 24 | |
| 25 # include "incls/_precompiled.incl" | |
| 26 # include "incls/_c1_IR.cpp.incl" | |
| 27 | |
| 28 | |
| 29 // Implementation of XHandlers | |
| 30 // | |
| 31 // Note: This code could eventually go away if we are | |
| 32 // just using the ciExceptionHandlerStream. | |
| 33 | |
| 34 XHandlers::XHandlers(ciMethod* method) : _list(method->exception_table_length()) { | |
| 35 ciExceptionHandlerStream s(method); | |
| 36 while (!s.is_done()) { | |
| 37 _list.append(new XHandler(s.handler())); | |
| 38 s.next(); | |
| 39 } | |
| 40 assert(s.count() == method->exception_table_length(), "exception table lengths inconsistent"); | |
| 41 } | |
| 42 | |
| 43 // deep copy of all XHandler contained in list | |
| 44 XHandlers::XHandlers(XHandlers* other) : | |
| 45 _list(other->length()) | |
| 46 { | |
| 47 for (int i = 0; i < other->length(); i++) { | |
| 48 _list.append(new XHandler(other->handler_at(i))); | |
| 49 } | |
| 50 } | |
| 51 | |
| 52 // Returns whether a particular exception type can be caught. Also | |
| 53 // returns true if klass is unloaded or any exception handler | |
| 54 // classes are unloaded. type_is_exact indicates whether the throw | |
| 55 // is known to be exactly that class or it might throw a subtype. | |
| 56 bool XHandlers::could_catch(ciInstanceKlass* klass, bool type_is_exact) const { | |
| 57 // the type is unknown so be conservative | |
| 58 if (!klass->is_loaded()) { | |
| 59 return true; | |
| 60 } | |
| 61 | |
| 62 for (int i = 0; i < length(); i++) { | |
| 63 XHandler* handler = handler_at(i); | |
| 64 if (handler->is_catch_all()) { | |
| 65 // catch of ANY | |
| 66 return true; | |
| 67 } | |
| 68 ciInstanceKlass* handler_klass = handler->catch_klass(); | |
| 69 // if it's unknown it might be catchable | |
| 70 if (!handler_klass->is_loaded()) { | |
| 71 return true; | |
| 72 } | |
| 73 // if the throw type is definitely a subtype of the catch type | |
| 74 // then it can be caught. | |
| 75 if (klass->is_subtype_of(handler_klass)) { | |
| 76 return true; | |
| 77 } | |
| 78 if (!type_is_exact) { | |
| 79 // If the type isn't exactly known then it can also be caught by | |
| 80 // catch statements where the inexact type is a subtype of the | |
| 81 // catch type. | |
| 82 // given: foo extends bar extends Exception | |
| 83 // throw bar can be caught by catch foo, catch bar, and catch | |
| 84 // Exception, however it can't be caught by any handlers without | |
| 85 // bar in its type hierarchy. | |
| 86 if (handler_klass->is_subtype_of(klass)) { | |
| 87 return true; | |
| 88 } | |
| 89 } | |
| 90 } | |
| 91 | |
| 92 return false; | |
| 93 } | |
| 94 | |
| 95 | |
| 96 bool XHandlers::equals(XHandlers* others) const { | |
| 97 if (others == NULL) return false; | |
| 98 if (length() != others->length()) return false; | |
| 99 | |
| 100 for (int i = 0; i < length(); i++) { | |
| 101 if (!handler_at(i)->equals(others->handler_at(i))) return false; | |
| 102 } | |
| 103 return true; | |
| 104 } | |
| 105 | |
| 106 bool XHandler::equals(XHandler* other) const { | |
| 107 assert(entry_pco() != -1 && other->entry_pco() != -1, "must have entry_pco"); | |
| 108 | |
| 109 if (entry_pco() != other->entry_pco()) return false; | |
| 110 if (scope_count() != other->scope_count()) return false; | |
| 111 if (_desc != other->_desc) return false; | |
| 112 | |
| 113 assert(entry_block() == other->entry_block(), "entry_block must be equal when entry_pco is equal"); | |
| 114 return true; | |
| 115 } | |
| 116 | |
| 117 | |
| 118 // Implementation of IRScope | |
| 119 | |
| 120 BlockBegin* IRScope::header_block(BlockBegin* entry, BlockBegin::Flag f, ValueStack* state) { | |
| 121 if (entry == NULL) return NULL; | |
| 122 assert(entry->is_set(f), "entry/flag mismatch"); | |
| 123 // create header block | |
| 124 BlockBegin* h = new BlockBegin(entry->bci()); | |
| 125 BlockEnd* g = new Goto(entry, false); | |
| 126 h->set_next(g, entry->bci()); | |
| 127 h->set_end(g); | |
| 128 h->set(f); | |
| 129 // setup header block end state | |
| 130 ValueStack* s = state->copy(); // can use copy since stack is empty (=> no phis) | |
| 131 assert(s->stack_is_empty(), "must have empty stack at entry point"); | |
| 132 g->set_state(s); | |
| 133 return h; | |
| 134 } | |
| 135 | |
| 136 | |
| 137 BlockBegin* IRScope::build_graph(Compilation* compilation, int osr_bci) { | |
| 138 GraphBuilder gm(compilation, this); | |
| 139 NOT_PRODUCT(if (PrintValueNumbering && Verbose) gm.print_stats()); | |
| 140 if (compilation->bailed_out()) return NULL; | |
| 141 return gm.start(); | |
| 142 } | |
| 143 | |
| 144 | |
| 145 IRScope::IRScope(Compilation* compilation, IRScope* caller, int caller_bci, ciMethod* method, int osr_bci, bool create_graph) | |
| 146 : _callees(2) | |
| 147 , _compilation(compilation) | |
| 148 , _lock_stack_size(-1) | |
| 149 , _requires_phi_function(method->max_locals()) | |
| 150 { | |
| 151 _caller = caller; | |
| 152 _caller_bci = caller == NULL ? -1 : caller_bci; | |
| 153 _caller_state = NULL; // Must be set later if needed | |
| 154 _level = caller == NULL ? 0 : caller->level() + 1; | |
| 155 _method = method; | |
| 156 _xhandlers = new XHandlers(method); | |
| 157 _number_of_locks = 0; | |
| 158 _monitor_pairing_ok = method->has_balanced_monitors(); | |
| 159 _start = NULL; | |
| 160 | |
| 161 if (osr_bci == -1) { | |
| 162 _requires_phi_function.clear(); | |
| 163 } else { | |
| 164 // selective creation of phi functions is not possibel in osr-methods | |
| 165 _requires_phi_function.set_range(0, method->max_locals()); | |
| 166 } | |
| 167 | |
| 168 assert(method->holder()->is_loaded() , "method holder must be loaded"); | |
| 169 | |
| 170 // build graph if monitor pairing is ok | |
| 171 if (create_graph && monitor_pairing_ok()) _start = build_graph(compilation, osr_bci); | |
| 172 } | |
| 173 | |
| 174 | |
| 175 int IRScope::max_stack() const { | |
| 176 int my_max = method()->max_stack(); | |
| 177 int callee_max = 0; | |
| 178 for (int i = 0; i < number_of_callees(); i++) { | |
| 179 callee_max = MAX2(callee_max, callee_no(i)->max_stack()); | |
| 180 } | |
| 181 return my_max + callee_max; | |
| 182 } | |
| 183 | |
| 184 | |
| 185 void IRScope::compute_lock_stack_size() { | |
| 186 if (!InlineMethodsWithExceptionHandlers) { | |
| 187 _lock_stack_size = 0; | |
| 188 return; | |
| 189 } | |
| 190 | |
| 191 // Figure out whether we have to preserve expression stack elements | |
| 192 // for parent scopes, and if so, how many | |
| 193 IRScope* cur_scope = this; | |
| 194 while (cur_scope != NULL && !cur_scope->xhandlers()->has_handlers()) { | |
| 195 cur_scope = cur_scope->caller(); | |
| 196 } | |
| 197 _lock_stack_size = (cur_scope == NULL ? 0 : | |
| 198 (cur_scope->caller_state() == NULL ? 0 : | |
| 199 cur_scope->caller_state()->stack_size())); | |
| 200 } | |
| 201 | |
| 202 int IRScope::top_scope_bci() const { | |
| 203 assert(!is_top_scope(), "no correct answer for top scope possible"); | |
| 204 const IRScope* scope = this; | |
| 205 while (!scope->caller()->is_top_scope()) { | |
| 206 scope = scope->caller(); | |
| 207 } | |
| 208 return scope->caller_bci(); | |
| 209 } | |
| 210 | |
|
900
9987d9d5eb0e
6833129: specjvm98 fails with NullPointerException in the compiler with -XX:DeoptimizeALot
cfang
parents:
470
diff
changeset
|
211 bool IRScopeDebugInfo::should_reexecute() { |
|
9987d9d5eb0e
6833129: specjvm98 fails with NullPointerException in the compiler with -XX:DeoptimizeALot
cfang
parents:
470
diff
changeset
|
212 ciMethod* cur_method = scope()->method(); |
|
9987d9d5eb0e
6833129: specjvm98 fails with NullPointerException in the compiler with -XX:DeoptimizeALot
cfang
parents:
470
diff
changeset
|
213 int cur_bci = bci(); |
|
9987d9d5eb0e
6833129: specjvm98 fails with NullPointerException in the compiler with -XX:DeoptimizeALot
cfang
parents:
470
diff
changeset
|
214 if (cur_method != NULL && cur_bci != SynchronizationEntryBCI) { |
|
9987d9d5eb0e
6833129: specjvm98 fails with NullPointerException in the compiler with -XX:DeoptimizeALot
cfang
parents:
470
diff
changeset
|
215 Bytecodes::Code code = cur_method->java_code_at_bci(cur_bci); |
|
9987d9d5eb0e
6833129: specjvm98 fails with NullPointerException in the compiler with -XX:DeoptimizeALot
cfang
parents:
470
diff
changeset
|
216 return Interpreter::bytecode_should_reexecute(code); |
|
9987d9d5eb0e
6833129: specjvm98 fails with NullPointerException in the compiler with -XX:DeoptimizeALot
cfang
parents:
470
diff
changeset
|
217 } else |
|
9987d9d5eb0e
6833129: specjvm98 fails with NullPointerException in the compiler with -XX:DeoptimizeALot
cfang
parents:
470
diff
changeset
|
218 return false; |
|
9987d9d5eb0e
6833129: specjvm98 fails with NullPointerException in the compiler with -XX:DeoptimizeALot
cfang
parents:
470
diff
changeset
|
219 } |
| 0 | 220 |
| 221 | |
| 222 // Implementation of CodeEmitInfo | |
| 223 | |
| 224 // Stack must be NON-null | |
| 225 CodeEmitInfo::CodeEmitInfo(int bci, ValueStack* stack, XHandlers* exception_handlers) | |
| 226 : _scope(stack->scope()) | |
| 227 , _bci(bci) | |
| 228 , _scope_debug_info(NULL) | |
| 229 , _oop_map(NULL) | |
| 230 , _stack(stack) | |
| 231 , _exception_handlers(exception_handlers) | |
| 232 , _next(NULL) | |
| 1564 | 233 , _id(-1) |
| 234 , _is_method_handle_invoke(false) { | |
| 0 | 235 assert(_stack != NULL, "must be non null"); |
| 236 assert(_bci == SynchronizationEntryBCI || Bytecodes::is_defined(scope()->method()->java_code_at_bci(_bci)), "make sure bci points at a real bytecode"); | |
| 237 } | |
| 238 | |
| 239 | |
| 240 CodeEmitInfo::CodeEmitInfo(CodeEmitInfo* info, bool lock_stack_only) | |
| 241 : _scope(info->_scope) | |
| 242 , _exception_handlers(NULL) | |
| 243 , _bci(info->_bci) | |
| 244 , _scope_debug_info(NULL) | |
| 1564 | 245 , _oop_map(NULL) |
| 246 , _is_method_handle_invoke(info->_is_method_handle_invoke) { | |
| 0 | 247 if (lock_stack_only) { |
| 248 if (info->_stack != NULL) { | |
| 249 _stack = info->_stack->copy_locks(); | |
| 250 } else { | |
| 251 _stack = NULL; | |
| 252 } | |
| 253 } else { | |
| 254 _stack = info->_stack; | |
| 255 } | |
| 256 | |
| 257 // deep copy of exception handlers | |
| 258 if (info->_exception_handlers != NULL) { | |
| 259 _exception_handlers = new XHandlers(info->_exception_handlers); | |
| 260 } | |
| 261 } | |
| 262 | |
| 263 | |
| 1564 | 264 void CodeEmitInfo::record_debug_info(DebugInformationRecorder* recorder, int pc_offset) { |
| 0 | 265 // record the safepoint before recording the debug info for enclosing scopes |
| 266 recorder->add_safepoint(pc_offset, _oop_map->deep_copy()); | |
| 1564 | 267 _scope_debug_info->record_debug_info(recorder, pc_offset, true/*topmost*/, _is_method_handle_invoke); |
| 0 | 268 recorder->end_safepoint(pc_offset); |
| 269 } | |
| 270 | |
| 271 | |
| 272 void CodeEmitInfo::add_register_oop(LIR_Opr opr) { | |
| 273 assert(_oop_map != NULL, "oop map must already exist"); | |
| 274 assert(opr->is_single_cpu(), "should not call otherwise"); | |
| 275 | |
| 276 int frame_size = frame_map()->framesize(); | |
| 277 int arg_count = frame_map()->oop_map_arg_count(); | |
| 278 VMReg name = frame_map()->regname(opr); | |
| 279 _oop_map->set_oop(name); | |
| 280 } | |
| 281 | |
| 282 | |
| 283 | |
| 284 | |
| 285 // Implementation of IR | |
| 286 | |
| 287 IR::IR(Compilation* compilation, ciMethod* method, int osr_bci) : | |
| 288 _locals_size(in_WordSize(-1)) | |
| 289 , _num_loops(0) { | |
| 290 // initialize data structures | |
| 291 ValueType::initialize(); | |
| 292 Instruction::initialize(); | |
| 293 BlockBegin::initialize(); | |
| 294 GraphBuilder::initialize(); | |
| 295 // setup IR fields | |
| 296 _compilation = compilation; | |
| 297 _top_scope = new IRScope(compilation, NULL, -1, method, osr_bci, true); | |
| 298 _code = NULL; | |
| 299 } | |
| 300 | |
| 301 | |
| 302 void IR::optimize() { | |
| 303 Optimizer opt(this); | |
| 304 if (DoCEE) { | |
| 305 opt.eliminate_conditional_expressions(); | |
| 306 #ifndef PRODUCT | |
| 307 if (PrintCFG || PrintCFG1) { tty->print_cr("CFG after CEE"); print(true); } | |
| 308 if (PrintIR || PrintIR1 ) { tty->print_cr("IR after CEE"); print(false); } | |
| 309 #endif | |
| 310 } | |
| 311 if (EliminateBlocks) { | |
| 312 opt.eliminate_blocks(); | |
| 313 #ifndef PRODUCT | |
| 314 if (PrintCFG || PrintCFG1) { tty->print_cr("CFG after block elimination"); print(true); } | |
| 315 if (PrintIR || PrintIR1 ) { tty->print_cr("IR after block elimination"); print(false); } | |
| 316 #endif | |
| 317 } | |
| 318 if (EliminateNullChecks) { | |
| 319 opt.eliminate_null_checks(); | |
| 320 #ifndef PRODUCT | |
| 321 if (PrintCFG || PrintCFG1) { tty->print_cr("CFG after null check elimination"); print(true); } | |
| 322 if (PrintIR || PrintIR1 ) { tty->print_cr("IR after null check elimination"); print(false); } | |
| 323 #endif | |
| 324 } | |
| 325 } | |
| 326 | |
| 327 | |
| 328 static int sort_pairs(BlockPair** a, BlockPair** b) { | |
| 329 if ((*a)->from() == (*b)->from()) { | |
| 330 return (*a)->to()->block_id() - (*b)->to()->block_id(); | |
| 331 } else { | |
| 332 return (*a)->from()->block_id() - (*b)->from()->block_id(); | |
| 333 } | |
| 334 } | |
| 335 | |
| 336 | |
| 337 class CriticalEdgeFinder: public BlockClosure { | |
| 338 BlockPairList blocks; | |
| 339 IR* _ir; | |
| 340 | |
| 341 public: | |
| 342 CriticalEdgeFinder(IR* ir): _ir(ir) {} | |
| 343 void block_do(BlockBegin* bb) { | |
| 344 BlockEnd* be = bb->end(); | |
| 345 int nos = be->number_of_sux(); | |
| 346 if (nos >= 2) { | |
| 347 for (int i = 0; i < nos; i++) { | |
| 348 BlockBegin* sux = be->sux_at(i); | |
| 349 if (sux->number_of_preds() >= 2) { | |
| 350 blocks.append(new BlockPair(bb, sux)); | |
| 351 } | |
| 352 } | |
| 353 } | |
| 354 } | |
| 355 | |
| 356 void split_edges() { | |
| 357 BlockPair* last_pair = NULL; | |
| 358 blocks.sort(sort_pairs); | |
| 359 for (int i = 0; i < blocks.length(); i++) { | |
| 360 BlockPair* pair = blocks.at(i); | |
| 361 if (last_pair != NULL && pair->is_same(last_pair)) continue; | |
| 362 BlockBegin* from = pair->from(); | |
| 363 BlockBegin* to = pair->to(); | |
| 364 BlockBegin* split = from->insert_block_between(to); | |
| 365 #ifndef PRODUCT | |
| 366 if ((PrintIR || PrintIR1) && Verbose) { | |
| 367 tty->print_cr("Split critical edge B%d -> B%d (new block B%d)", | |
| 368 from->block_id(), to->block_id(), split->block_id()); | |
| 369 } | |
| 370 #endif | |
| 371 last_pair = pair; | |
| 372 } | |
| 373 } | |
| 374 }; | |
| 375 | |
| 376 void IR::split_critical_edges() { | |
| 377 CriticalEdgeFinder cef(this); | |
| 378 | |
| 379 iterate_preorder(&cef); | |
| 380 cef.split_edges(); | |
| 381 } | |
| 382 | |
| 383 | |
| 384 class UseCountComputer: public AllStatic { | |
| 385 private: | |
| 386 static void update_use_count(Value* n) { | |
| 387 // Local instructions and Phis for expression stack values at the | |
| 388 // start of basic blocks are not added to the instruction list | |
| 389 if ((*n)->bci() == -99 && (*n)->as_Local() == NULL && | |
| 390 (*n)->as_Phi() == NULL) { | |
| 391 assert(false, "a node was not appended to the graph"); | |
| 392 Compilation::current_compilation()->bailout("a node was not appended to the graph"); | |
| 393 } | |
| 394 // use n's input if not visited before | |
| 395 if (!(*n)->is_pinned() && !(*n)->has_uses()) { | |
| 396 // note: a) if the instruction is pinned, it will be handled by compute_use_count | |
| 397 // b) if the instruction has uses, it was touched before | |
| 398 // => in both cases we don't need to update n's values | |
| 399 uses_do(n); | |
| 400 } | |
| 401 // use n | |
| 402 (*n)->_use_count++; | |
| 403 } | |
| 404 | |
| 405 static Values* worklist; | |
| 406 static int depth; | |
| 407 enum { | |
| 408 max_recurse_depth = 20 | |
| 409 }; | |
| 410 | |
| 411 static void uses_do(Value* n) { | |
| 412 depth++; | |
| 413 if (depth > max_recurse_depth) { | |
| 414 // don't allow the traversal to recurse too deeply | |
| 415 worklist->push(*n); | |
| 416 } else { | |
| 417 (*n)->input_values_do(update_use_count); | |
| 418 // special handling for some instructions | |
| 419 if ((*n)->as_BlockEnd() != NULL) { | |
| 420 // note on BlockEnd: | |
| 421 // must 'use' the stack only if the method doesn't | |
| 422 // terminate, however, in those cases stack is empty | |
| 423 (*n)->state_values_do(update_use_count); | |
| 424 } | |
| 425 } | |
| 426 depth--; | |
| 427 } | |
| 428 | |
| 429 static void basic_compute_use_count(BlockBegin* b) { | |
| 430 depth = 0; | |
| 431 // process all pinned nodes as the roots of expression trees | |
| 432 for (Instruction* n = b; n != NULL; n = n->next()) { | |
| 433 if (n->is_pinned()) uses_do(&n); | |
| 434 } | |
| 435 assert(depth == 0, "should have counted back down"); | |
| 436 | |
| 437 // now process any unpinned nodes which recursed too deeply | |
| 438 while (worklist->length() > 0) { | |
| 439 Value t = worklist->pop(); | |
| 440 if (!t->is_pinned()) { | |
| 441 // compute the use count | |
| 442 uses_do(&t); | |
| 443 | |
| 444 // pin the instruction so that LIRGenerator doesn't recurse | |
| 445 // too deeply during it's evaluation. | |
| 446 t->pin(); | |
| 447 } | |
| 448 } | |
| 449 assert(depth == 0, "should have counted back down"); | |
| 450 } | |
| 451 | |
| 452 public: | |
| 453 static void compute(BlockList* blocks) { | |
| 454 worklist = new Values(); | |
| 455 blocks->blocks_do(basic_compute_use_count); | |
| 456 worklist = NULL; | |
| 457 } | |
| 458 }; | |
| 459 | |
| 460 | |
| 461 Values* UseCountComputer::worklist = NULL; | |
| 462 int UseCountComputer::depth = 0; | |
| 463 | |
| 464 // helper macro for short definition of trace-output inside code | |
| 465 #ifndef PRODUCT | |
| 466 #define TRACE_LINEAR_SCAN(level, code) \ | |
| 467 if (TraceLinearScanLevel >= level) { \ | |
| 468 code; \ | |
| 469 } | |
| 470 #else | |
| 471 #define TRACE_LINEAR_SCAN(level, code) | |
| 472 #endif | |
| 473 | |
| 474 class ComputeLinearScanOrder : public StackObj { | |
| 475 private: | |
| 476 int _max_block_id; // the highest block_id of a block | |
| 477 int _num_blocks; // total number of blocks (smaller than _max_block_id) | |
| 478 int _num_loops; // total number of loops | |
| 479 bool _iterative_dominators;// method requires iterative computation of dominatiors | |
| 480 | |
| 481 BlockList* _linear_scan_order; // the resulting list of blocks in correct order | |
| 482 | |
| 483 BitMap _visited_blocks; // used for recursive processing of blocks | |
| 484 BitMap _active_blocks; // used for recursive processing of blocks | |
| 485 BitMap _dominator_blocks; // temproary BitMap used for computation of dominator | |
| 486 intArray _forward_branches; // number of incoming forward branches for each block | |
| 487 BlockList _loop_end_blocks; // list of all loop end blocks collected during count_edges | |
| 488 BitMap2D _loop_map; // two-dimensional bit set: a bit is set if a block is contained in a loop | |
| 489 BlockList _work_list; // temporary list (used in mark_loops and compute_order) | |
| 490 | |
| 491 // accessors for _visited_blocks and _active_blocks | |
| 492 void init_visited() { _active_blocks.clear(); _visited_blocks.clear(); } | |
| 493 bool is_visited(BlockBegin* b) const { return _visited_blocks.at(b->block_id()); } | |
| 494 bool is_active(BlockBegin* b) const { return _active_blocks.at(b->block_id()); } | |
| 495 void set_visited(BlockBegin* b) { assert(!is_visited(b), "already set"); _visited_blocks.set_bit(b->block_id()); } | |
| 496 void set_active(BlockBegin* b) { assert(!is_active(b), "already set"); _active_blocks.set_bit(b->block_id()); } | |
| 497 void clear_active(BlockBegin* b) { assert(is_active(b), "not already"); _active_blocks.clear_bit(b->block_id()); } | |
| 498 | |
| 499 // accessors for _forward_branches | |
| 500 void inc_forward_branches(BlockBegin* b) { _forward_branches.at_put(b->block_id(), _forward_branches.at(b->block_id()) + 1); } | |
| 501 int dec_forward_branches(BlockBegin* b) { _forward_branches.at_put(b->block_id(), _forward_branches.at(b->block_id()) - 1); return _forward_branches.at(b->block_id()); } | |
| 502 | |
| 503 // accessors for _loop_map | |
| 504 bool is_block_in_loop (int loop_idx, BlockBegin* b) const { return _loop_map.at(loop_idx, b->block_id()); } | |
| 505 void set_block_in_loop (int loop_idx, BlockBegin* b) { _loop_map.set_bit(loop_idx, b->block_id()); } | |
| 506 void clear_block_in_loop(int loop_idx, int block_id) { _loop_map.clear_bit(loop_idx, block_id); } | |
| 507 | |
| 508 // count edges between blocks | |
| 509 void count_edges(BlockBegin* cur, BlockBegin* parent); | |
| 510 | |
| 511 // loop detection | |
| 512 void mark_loops(); | |
| 513 void clear_non_natural_loops(BlockBegin* start_block); | |
| 514 void assign_loop_depth(BlockBegin* start_block); | |
| 515 | |
| 516 // computation of final block order | |
| 517 BlockBegin* common_dominator(BlockBegin* a, BlockBegin* b); | |
| 518 void compute_dominator(BlockBegin* cur, BlockBegin* parent); | |
| 519 int compute_weight(BlockBegin* cur); | |
| 520 bool ready_for_processing(BlockBegin* cur); | |
| 521 void sort_into_work_list(BlockBegin* b); | |
| 522 void append_block(BlockBegin* cur); | |
| 523 void compute_order(BlockBegin* start_block); | |
| 524 | |
| 525 // fixup of dominators for non-natural loops | |
| 526 bool compute_dominators_iter(); | |
| 527 void compute_dominators(); | |
| 528 | |
| 529 // debug functions | |
| 530 NOT_PRODUCT(void print_blocks();) | |
| 531 DEBUG_ONLY(void verify();) | |
| 532 | |
| 533 public: | |
| 534 ComputeLinearScanOrder(BlockBegin* start_block); | |
| 535 | |
| 536 // accessors for final result | |
| 537 BlockList* linear_scan_order() const { return _linear_scan_order; } | |
| 538 int num_loops() const { return _num_loops; } | |
| 539 }; | |
| 540 | |
| 541 | |
| 542 ComputeLinearScanOrder::ComputeLinearScanOrder(BlockBegin* start_block) : | |
| 543 _max_block_id(BlockBegin::number_of_blocks()), | |
| 544 _num_blocks(0), | |
| 545 _num_loops(0), | |
| 546 _iterative_dominators(false), | |
| 547 _visited_blocks(_max_block_id), | |
| 548 _active_blocks(_max_block_id), | |
| 549 _dominator_blocks(_max_block_id), | |
| 550 _forward_branches(_max_block_id, 0), | |
| 551 _loop_end_blocks(8), | |
| 552 _work_list(8), | |
| 553 _linear_scan_order(NULL), // initialized later with correct size | |
| 554 _loop_map(0, 0) // initialized later with correct size | |
| 555 { | |
| 556 TRACE_LINEAR_SCAN(2, "***** computing linear-scan block order"); | |
| 557 | |
| 558 init_visited(); | |
| 559 count_edges(start_block, NULL); | |
| 560 | |
| 561 if (_num_loops > 0) { | |
| 562 mark_loops(); | |
| 563 clear_non_natural_loops(start_block); | |
| 564 assign_loop_depth(start_block); | |
| 565 } | |
| 566 | |
| 567 compute_order(start_block); | |
| 568 compute_dominators(); | |
| 569 | |
| 570 NOT_PRODUCT(print_blocks()); | |
| 571 DEBUG_ONLY(verify()); | |
| 572 } | |
| 573 | |
| 574 | |
| 575 // Traverse the CFG: | |
| 576 // * count total number of blocks | |
| 577 // * count all incoming edges and backward incoming edges | |
| 578 // * number loop header blocks | |
| 579 // * create a list with all loop end blocks | |
| 580 void ComputeLinearScanOrder::count_edges(BlockBegin* cur, BlockBegin* parent) { | |
| 581 TRACE_LINEAR_SCAN(3, tty->print_cr("Enter count_edges for block B%d coming from B%d", cur->block_id(), parent != NULL ? parent->block_id() : -1)); | |
| 582 assert(cur->dominator() == NULL, "dominator already initialized"); | |
| 583 | |
| 584 if (is_active(cur)) { | |
| 585 TRACE_LINEAR_SCAN(3, tty->print_cr("backward branch")); | |
| 586 assert(is_visited(cur), "block must be visisted when block is active"); | |
| 587 assert(parent != NULL, "must have parent"); | |
| 588 | |
| 589 cur->set(BlockBegin::linear_scan_loop_header_flag); | |
| 590 cur->set(BlockBegin::backward_branch_target_flag); | |
| 591 | |
| 592 parent->set(BlockBegin::linear_scan_loop_end_flag); | |
|
428
334969144810
6758445: loop heads that are exception entry points can crash during count_edges/mark_loops
never
parents:
0
diff
changeset
|
593 |
|
334969144810
6758445: loop heads that are exception entry points can crash during count_edges/mark_loops
never
parents:
0
diff
changeset
|
594 // When a loop header is also the start of an exception handler, then the backward branch is |
|
334969144810
6758445: loop heads that are exception entry points can crash during count_edges/mark_loops
never
parents:
0
diff
changeset
|
595 // an exception edge. Because such edges are usually critical edges which cannot be split, the |
|
334969144810
6758445: loop heads that are exception entry points can crash during count_edges/mark_loops
never
parents:
0
diff
changeset
|
596 // loop must be excluded here from processing. |
|
334969144810
6758445: loop heads that are exception entry points can crash during count_edges/mark_loops
never
parents:
0
diff
changeset
|
597 if (cur->is_set(BlockBegin::exception_entry_flag)) { |
|
334969144810
6758445: loop heads that are exception entry points can crash during count_edges/mark_loops
never
parents:
0
diff
changeset
|
598 // Make sure that dominators are correct in this weird situation |
|
334969144810
6758445: loop heads that are exception entry points can crash during count_edges/mark_loops
never
parents:
0
diff
changeset
|
599 _iterative_dominators = true; |
|
334969144810
6758445: loop heads that are exception entry points can crash during count_edges/mark_loops
never
parents:
0
diff
changeset
|
600 return; |
|
334969144810
6758445: loop heads that are exception entry points can crash during count_edges/mark_loops
never
parents:
0
diff
changeset
|
601 } |
|
334969144810
6758445: loop heads that are exception entry points can crash during count_edges/mark_loops
never
parents:
0
diff
changeset
|
602 assert(parent->number_of_sux() == 1 && parent->sux_at(0) == cur, |
|
334969144810
6758445: loop heads that are exception entry points can crash during count_edges/mark_loops
never
parents:
0
diff
changeset
|
603 "loop end blocks must have one successor (critical edges are split)"); |
|
334969144810
6758445: loop heads that are exception entry points can crash during count_edges/mark_loops
never
parents:
0
diff
changeset
|
604 |
| 0 | 605 _loop_end_blocks.append(parent); |
| 606 return; | |
| 607 } | |
| 608 | |
| 609 // increment number of incoming forward branches | |
| 610 inc_forward_branches(cur); | |
| 611 | |
| 612 if (is_visited(cur)) { | |
| 613 TRACE_LINEAR_SCAN(3, tty->print_cr("block already visited")); | |
| 614 return; | |
| 615 } | |
| 616 | |
| 617 _num_blocks++; | |
| 618 set_visited(cur); | |
| 619 set_active(cur); | |
| 620 | |
| 621 // recursive call for all successors | |
| 622 int i; | |
| 623 for (i = cur->number_of_sux() - 1; i >= 0; i--) { | |
| 624 count_edges(cur->sux_at(i), cur); | |
| 625 } | |
| 626 for (i = cur->number_of_exception_handlers() - 1; i >= 0; i--) { | |
| 627 count_edges(cur->exception_handler_at(i), cur); | |
| 628 } | |
| 629 | |
| 630 clear_active(cur); | |
| 631 | |
| 632 // Each loop has a unique number. | |
| 633 // When multiple loops are nested, assign_loop_depth assumes that the | |
| 634 // innermost loop has the lowest number. This is guaranteed by setting | |
| 635 // the loop number after the recursive calls for the successors above | |
| 636 // have returned. | |
| 637 if (cur->is_set(BlockBegin::linear_scan_loop_header_flag)) { | |
| 638 assert(cur->loop_index() == -1, "cannot set loop-index twice"); | |
| 639 TRACE_LINEAR_SCAN(3, tty->print_cr("Block B%d is loop header of loop %d", cur->block_id(), _num_loops)); | |
| 640 | |
| 641 cur->set_loop_index(_num_loops); | |
| 642 _num_loops++; | |
| 643 } | |
| 644 | |
| 645 TRACE_LINEAR_SCAN(3, tty->print_cr("Finished count_edges for block B%d", cur->block_id())); | |
| 646 } | |
| 647 | |
| 648 | |
| 649 void ComputeLinearScanOrder::mark_loops() { | |
| 650 TRACE_LINEAR_SCAN(3, tty->print_cr("----- marking loops")); | |
| 651 | |
| 652 _loop_map = BitMap2D(_num_loops, _max_block_id); | |
| 653 _loop_map.clear(); | |
| 654 | |
| 655 for (int i = _loop_end_blocks.length() - 1; i >= 0; i--) { | |
| 656 BlockBegin* loop_end = _loop_end_blocks.at(i); | |
| 657 BlockBegin* loop_start = loop_end->sux_at(0); | |
| 658 int loop_idx = loop_start->loop_index(); | |
| 659 | |
| 660 TRACE_LINEAR_SCAN(3, tty->print_cr("Processing loop from B%d to B%d (loop %d):", loop_start->block_id(), loop_end->block_id(), loop_idx)); | |
| 661 assert(loop_end->is_set(BlockBegin::linear_scan_loop_end_flag), "loop end flag must be set"); | |
| 662 assert(loop_end->number_of_sux() == 1, "incorrect number of successors"); | |
| 663 assert(loop_start->is_set(BlockBegin::linear_scan_loop_header_flag), "loop header flag must be set"); | |
| 664 assert(loop_idx >= 0 && loop_idx < _num_loops, "loop index not set"); | |
| 665 assert(_work_list.is_empty(), "work list must be empty before processing"); | |
| 666 | |
| 667 // add the end-block of the loop to the working list | |
| 668 _work_list.push(loop_end); | |
| 669 set_block_in_loop(loop_idx, loop_end); | |
| 670 do { | |
| 671 BlockBegin* cur = _work_list.pop(); | |
| 672 | |
| 673 TRACE_LINEAR_SCAN(3, tty->print_cr(" processing B%d", cur->block_id())); | |
| 674 assert(is_block_in_loop(loop_idx, cur), "bit in loop map must be set when block is in work list"); | |
| 675 | |
| 676 // recursive processing of all predecessors ends when start block of loop is reached | |
| 677 if (cur != loop_start && !cur->is_set(BlockBegin::osr_entry_flag)) { | |
| 678 for (int j = cur->number_of_preds() - 1; j >= 0; j--) { | |
| 679 BlockBegin* pred = cur->pred_at(j); | |
| 680 | |
| 681 if (!is_block_in_loop(loop_idx, pred) /*&& !pred->is_set(BlockBeginosr_entry_flag)*/) { | |
| 682 // this predecessor has not been processed yet, so add it to work list | |
| 683 TRACE_LINEAR_SCAN(3, tty->print_cr(" pushing B%d", pred->block_id())); | |
| 684 _work_list.push(pred); | |
| 685 set_block_in_loop(loop_idx, pred); | |
| 686 } | |
| 687 } | |
| 688 } | |
| 689 } while (!_work_list.is_empty()); | |
| 690 } | |
| 691 } | |
| 692 | |
| 693 | |
| 694 // check for non-natural loops (loops where the loop header does not dominate | |
| 695 // all other loop blocks = loops with mulitple entries). | |
| 696 // such loops are ignored | |
| 697 void ComputeLinearScanOrder::clear_non_natural_loops(BlockBegin* start_block) { | |
| 698 for (int i = _num_loops - 1; i >= 0; i--) { | |
| 699 if (is_block_in_loop(i, start_block)) { | |
| 700 // loop i contains the entry block of the method | |
| 701 // -> this is not a natural loop, so ignore it | |
| 702 TRACE_LINEAR_SCAN(2, tty->print_cr("Loop %d is non-natural, so it is ignored", i)); | |
| 703 | |
| 704 for (int block_id = _max_block_id - 1; block_id >= 0; block_id--) { | |
| 705 clear_block_in_loop(i, block_id); | |
| 706 } | |
| 707 _iterative_dominators = true; | |
| 708 } | |
| 709 } | |
| 710 } | |
| 711 | |
| 712 void ComputeLinearScanOrder::assign_loop_depth(BlockBegin* start_block) { | |
| 713 TRACE_LINEAR_SCAN(3, "----- computing loop-depth and weight"); | |
| 714 init_visited(); | |
| 715 | |
| 716 assert(_work_list.is_empty(), "work list must be empty before processing"); | |
| 717 _work_list.append(start_block); | |
| 718 | |
| 719 do { | |
| 720 BlockBegin* cur = _work_list.pop(); | |
| 721 | |
| 722 if (!is_visited(cur)) { | |
| 723 set_visited(cur); | |
| 724 TRACE_LINEAR_SCAN(4, tty->print_cr("Computing loop depth for block B%d", cur->block_id())); | |
| 725 | |
| 726 // compute loop-depth and loop-index for the block | |
| 727 assert(cur->loop_depth() == 0, "cannot set loop-depth twice"); | |
| 728 int i; | |
| 729 int loop_depth = 0; | |
| 730 int min_loop_idx = -1; | |
| 731 for (i = _num_loops - 1; i >= 0; i--) { | |
| 732 if (is_block_in_loop(i, cur)) { | |
| 733 loop_depth++; | |
| 734 min_loop_idx = i; | |
| 735 } | |
| 736 } | |
| 737 cur->set_loop_depth(loop_depth); | |
| 738 cur->set_loop_index(min_loop_idx); | |
| 739 | |
| 740 // append all unvisited successors to work list | |
| 741 for (i = cur->number_of_sux() - 1; i >= 0; i--) { | |
| 742 _work_list.append(cur->sux_at(i)); | |
| 743 } | |
| 744 for (i = cur->number_of_exception_handlers() - 1; i >= 0; i--) { | |
| 745 _work_list.append(cur->exception_handler_at(i)); | |
| 746 } | |
| 747 } | |
| 748 } while (!_work_list.is_empty()); | |
| 749 } | |
| 750 | |
| 751 | |
| 752 BlockBegin* ComputeLinearScanOrder::common_dominator(BlockBegin* a, BlockBegin* b) { | |
| 753 assert(a != NULL && b != NULL, "must have input blocks"); | |
| 754 | |
| 755 _dominator_blocks.clear(); | |
| 756 while (a != NULL) { | |
| 757 _dominator_blocks.set_bit(a->block_id()); | |
| 758 assert(a->dominator() != NULL || a == _linear_scan_order->at(0), "dominator must be initialized"); | |
| 759 a = a->dominator(); | |
| 760 } | |
| 761 while (b != NULL && !_dominator_blocks.at(b->block_id())) { | |
| 762 assert(b->dominator() != NULL || b == _linear_scan_order->at(0), "dominator must be initialized"); | |
| 763 b = b->dominator(); | |
| 764 } | |
| 765 | |
| 766 assert(b != NULL, "could not find dominator"); | |
| 767 return b; | |
| 768 } | |
| 769 | |
| 770 void ComputeLinearScanOrder::compute_dominator(BlockBegin* cur, BlockBegin* parent) { | |
| 771 if (cur->dominator() == NULL) { | |
| 772 TRACE_LINEAR_SCAN(4, tty->print_cr("DOM: initializing dominator of B%d to B%d", cur->block_id(), parent->block_id())); | |
| 773 cur->set_dominator(parent); | |
| 774 | |
| 775 } else if (!(cur->is_set(BlockBegin::linear_scan_loop_header_flag) && parent->is_set(BlockBegin::linear_scan_loop_end_flag))) { | |
| 776 TRACE_LINEAR_SCAN(4, tty->print_cr("DOM: computing dominator of B%d: common dominator of B%d and B%d is B%d", cur->block_id(), parent->block_id(), cur->dominator()->block_id(), common_dominator(cur->dominator(), parent)->block_id())); | |
| 777 assert(cur->number_of_preds() > 1, ""); | |
| 778 cur->set_dominator(common_dominator(cur->dominator(), parent)); | |
| 779 } | |
| 780 } | |
| 781 | |
| 782 | |
| 783 int ComputeLinearScanOrder::compute_weight(BlockBegin* cur) { | |
| 784 BlockBegin* single_sux = NULL; | |
| 785 if (cur->number_of_sux() == 1) { | |
| 786 single_sux = cur->sux_at(0); | |
| 787 } | |
| 788 | |
| 789 // limit loop-depth to 15 bit (only for security reason, it will never be so big) | |
| 790 int weight = (cur->loop_depth() & 0x7FFF) << 16; | |
| 791 | |
| 792 // general macro for short definition of weight flags | |
| 793 // the first instance of INC_WEIGHT_IF has the highest priority | |
| 794 int cur_bit = 15; | |
| 795 #define INC_WEIGHT_IF(condition) if ((condition)) { weight |= (1 << cur_bit); } cur_bit--; | |
| 796 | |
| 797 // this is necessery for the (very rare) case that two successing blocks have | |
| 798 // the same loop depth, but a different loop index (can happen for endless loops | |
| 799 // with exception handlers) | |
| 800 INC_WEIGHT_IF(!cur->is_set(BlockBegin::linear_scan_loop_header_flag)); | |
| 801 | |
| 802 // loop end blocks (blocks that end with a backward branch) are added | |
| 803 // after all other blocks of the loop. | |
| 804 INC_WEIGHT_IF(!cur->is_set(BlockBegin::linear_scan_loop_end_flag)); | |
| 805 | |
| 806 // critical edge split blocks are prefered because than they have a bigger | |
| 807 // proability to be completely empty | |
| 808 INC_WEIGHT_IF(cur->is_set(BlockBegin::critical_edge_split_flag)); | |
| 809 | |
| 810 // exceptions should not be thrown in normal control flow, so these blocks | |
| 811 // are added as late as possible | |
| 812 INC_WEIGHT_IF(cur->end()->as_Throw() == NULL && (single_sux == NULL || single_sux->end()->as_Throw() == NULL)); | |
| 813 INC_WEIGHT_IF(cur->end()->as_Return() == NULL && (single_sux == NULL || single_sux->end()->as_Return() == NULL)); | |
| 814 | |
| 815 // exceptions handlers are added as late as possible | |
| 816 INC_WEIGHT_IF(!cur->is_set(BlockBegin::exception_entry_flag)); | |
| 817 | |
| 818 // guarantee that weight is > 0 | |
| 819 weight |= 1; | |
| 820 | |
| 821 #undef INC_WEIGHT_IF | |
| 822 assert(cur_bit >= 0, "too many flags"); | |
| 823 assert(weight > 0, "weight cannot become negative"); | |
| 824 | |
| 825 return weight; | |
| 826 } | |
| 827 | |
| 828 bool ComputeLinearScanOrder::ready_for_processing(BlockBegin* cur) { | |
| 829 // Discount the edge just traveled. | |
| 830 // When the number drops to zero, all forward branches were processed | |
| 831 if (dec_forward_branches(cur) != 0) { | |
| 832 return false; | |
| 833 } | |
| 834 | |
| 835 assert(_linear_scan_order->index_of(cur) == -1, "block already processed (block can be ready only once)"); | |
| 836 assert(_work_list.index_of(cur) == -1, "block already in work-list (block can be ready only once)"); | |
| 837 return true; | |
| 838 } | |
| 839 | |
| 840 void ComputeLinearScanOrder::sort_into_work_list(BlockBegin* cur) { | |
| 841 assert(_work_list.index_of(cur) == -1, "block already in work list"); | |
| 842 | |
| 843 int cur_weight = compute_weight(cur); | |
| 844 | |
| 845 // the linear_scan_number is used to cache the weight of a block | |
| 846 cur->set_linear_scan_number(cur_weight); | |
| 847 | |
| 848 #ifndef PRODUCT | |
| 849 if (StressLinearScan) { | |
| 850 _work_list.insert_before(0, cur); | |
| 851 return; | |
| 852 } | |
| 853 #endif | |
| 854 | |
| 855 _work_list.append(NULL); // provide space for new element | |
| 856 | |
| 857 int insert_idx = _work_list.length() - 1; | |
| 858 while (insert_idx > 0 && _work_list.at(insert_idx - 1)->linear_scan_number() > cur_weight) { | |
| 859 _work_list.at_put(insert_idx, _work_list.at(insert_idx - 1)); | |
| 860 insert_idx--; | |
| 861 } | |
| 862 _work_list.at_put(insert_idx, cur); | |
| 863 | |
| 864 TRACE_LINEAR_SCAN(3, tty->print_cr("Sorted B%d into worklist. new worklist:", cur->block_id())); | |
| 865 TRACE_LINEAR_SCAN(3, for (int i = 0; i < _work_list.length(); i++) tty->print_cr("%8d B%2d weight:%6x", i, _work_list.at(i)->block_id(), _work_list.at(i)->linear_scan_number())); | |
| 866 | |
| 867 #ifdef ASSERT | |
| 868 for (int i = 0; i < _work_list.length(); i++) { | |
| 869 assert(_work_list.at(i)->linear_scan_number() > 0, "weight not set"); | |
| 870 assert(i == 0 || _work_list.at(i - 1)->linear_scan_number() <= _work_list.at(i)->linear_scan_number(), "incorrect order in worklist"); | |
| 871 } | |
| 872 #endif | |
| 873 } | |
| 874 | |
| 875 void ComputeLinearScanOrder::append_block(BlockBegin* cur) { | |
| 876 TRACE_LINEAR_SCAN(3, tty->print_cr("appending block B%d (weight 0x%6x) to linear-scan order", cur->block_id(), cur->linear_scan_number())); | |
| 877 assert(_linear_scan_order->index_of(cur) == -1, "cannot add the same block twice"); | |
| 878 | |
| 879 // currently, the linear scan order and code emit order are equal. | |
| 880 // therefore the linear_scan_number and the weight of a block must also | |
| 881 // be equal. | |
| 882 cur->set_linear_scan_number(_linear_scan_order->length()); | |
| 883 _linear_scan_order->append(cur); | |
| 884 } | |
| 885 | |
| 886 void ComputeLinearScanOrder::compute_order(BlockBegin* start_block) { | |
| 887 TRACE_LINEAR_SCAN(3, "----- computing final block order"); | |
| 888 | |
| 889 // the start block is always the first block in the linear scan order | |
| 890 _linear_scan_order = new BlockList(_num_blocks); | |
| 891 append_block(start_block); | |
| 892 | |
| 893 assert(start_block->end()->as_Base() != NULL, "start block must end with Base-instruction"); | |
| 894 BlockBegin* std_entry = ((Base*)start_block->end())->std_entry(); | |
| 895 BlockBegin* osr_entry = ((Base*)start_block->end())->osr_entry(); | |
| 896 | |
| 897 BlockBegin* sux_of_osr_entry = NULL; | |
| 898 if (osr_entry != NULL) { | |
| 899 // special handling for osr entry: | |
| 900 // ignore the edge between the osr entry and its successor for processing | |
| 901 // the osr entry block is added manually below | |
| 902 assert(osr_entry->number_of_sux() == 1, "osr entry must have exactly one successor"); | |
| 903 assert(osr_entry->sux_at(0)->number_of_preds() >= 2, "sucessor of osr entry must have two predecessors (otherwise it is not present in normal control flow"); | |
| 904 | |
| 905 sux_of_osr_entry = osr_entry->sux_at(0); | |
| 906 dec_forward_branches(sux_of_osr_entry); | |
| 907 | |
| 908 compute_dominator(osr_entry, start_block); | |
| 909 _iterative_dominators = true; | |
| 910 } | |
| 911 compute_dominator(std_entry, start_block); | |
| 912 | |
| 913 // start processing with standard entry block | |
| 914 assert(_work_list.is_empty(), "list must be empty before processing"); | |
| 915 | |
| 916 if (ready_for_processing(std_entry)) { | |
| 917 sort_into_work_list(std_entry); | |
| 918 } else { | |
| 919 assert(false, "the std_entry must be ready for processing (otherwise, the method has no start block)"); | |
| 920 } | |
| 921 | |
| 922 do { | |
| 923 BlockBegin* cur = _work_list.pop(); | |
| 924 | |
| 925 if (cur == sux_of_osr_entry) { | |
| 926 // the osr entry block is ignored in normal processing, it is never added to the | |
| 927 // work list. Instead, it is added as late as possible manually here. | |
| 928 append_block(osr_entry); | |
| 929 compute_dominator(cur, osr_entry); | |
| 930 } | |
| 931 append_block(cur); | |
| 932 | |
| 933 int i; | |
| 934 int num_sux = cur->number_of_sux(); | |
| 935 // changed loop order to get "intuitive" order of if- and else-blocks | |
| 936 for (i = 0; i < num_sux; i++) { | |
| 937 BlockBegin* sux = cur->sux_at(i); | |
| 938 compute_dominator(sux, cur); | |
| 939 if (ready_for_processing(sux)) { | |
| 940 sort_into_work_list(sux); | |
| 941 } | |
| 942 } | |
| 943 num_sux = cur->number_of_exception_handlers(); | |
| 944 for (i = 0; i < num_sux; i++) { | |
| 945 BlockBegin* sux = cur->exception_handler_at(i); | |
| 946 compute_dominator(sux, cur); | |
| 947 if (ready_for_processing(sux)) { | |
| 948 sort_into_work_list(sux); | |
| 949 } | |
| 950 } | |
| 951 } while (_work_list.length() > 0); | |
| 952 } | |
| 953 | |
| 954 | |
| 955 bool ComputeLinearScanOrder::compute_dominators_iter() { | |
| 956 bool changed = false; | |
| 957 int num_blocks = _linear_scan_order->length(); | |
| 958 | |
| 959 assert(_linear_scan_order->at(0)->dominator() == NULL, "must not have dominator"); | |
| 960 assert(_linear_scan_order->at(0)->number_of_preds() == 0, "must not have predecessors"); | |
| 961 for (int i = 1; i < num_blocks; i++) { | |
| 962 BlockBegin* block = _linear_scan_order->at(i); | |
| 963 | |
| 964 BlockBegin* dominator = block->pred_at(0); | |
| 965 int num_preds = block->number_of_preds(); | |
| 966 for (int i = 1; i < num_preds; i++) { | |
| 967 dominator = common_dominator(dominator, block->pred_at(i)); | |
| 968 } | |
| 969 | |
| 970 if (dominator != block->dominator()) { | |
| 971 TRACE_LINEAR_SCAN(4, tty->print_cr("DOM: updating dominator of B%d from B%d to B%d", block->block_id(), block->dominator()->block_id(), dominator->block_id())); | |
| 972 | |
| 973 block->set_dominator(dominator); | |
| 974 changed = true; | |
| 975 } | |
| 976 } | |
| 977 return changed; | |
| 978 } | |
| 979 | |
| 980 void ComputeLinearScanOrder::compute_dominators() { | |
| 981 TRACE_LINEAR_SCAN(3, tty->print_cr("----- computing dominators (iterative computation reqired: %d)", _iterative_dominators)); | |
| 982 | |
| 983 // iterative computation of dominators is only required for methods with non-natural loops | |
| 984 // and OSR-methods. For all other methods, the dominators computed when generating the | |
| 985 // linear scan block order are correct. | |
| 986 if (_iterative_dominators) { | |
| 987 do { | |
| 988 TRACE_LINEAR_SCAN(1, tty->print_cr("DOM: next iteration of fix-point calculation")); | |
| 989 } while (compute_dominators_iter()); | |
| 990 } | |
| 991 | |
| 992 // check that dominators are correct | |
| 993 assert(!compute_dominators_iter(), "fix point not reached"); | |
| 994 } | |
| 995 | |
| 996 | |
| 997 #ifndef PRODUCT | |
| 998 void ComputeLinearScanOrder::print_blocks() { | |
| 999 if (TraceLinearScanLevel >= 2) { | |
| 1000 tty->print_cr("----- loop information:"); | |
| 1001 for (int block_idx = 0; block_idx < _linear_scan_order->length(); block_idx++) { | |
| 1002 BlockBegin* cur = _linear_scan_order->at(block_idx); | |
| 1003 | |
| 1004 tty->print("%4d: B%2d: ", cur->linear_scan_number(), cur->block_id()); | |
| 1005 for (int loop_idx = 0; loop_idx < _num_loops; loop_idx++) { | |
| 1006 tty->print ("%d ", is_block_in_loop(loop_idx, cur)); | |
| 1007 } | |
| 1008 tty->print_cr(" -> loop_index: %2d, loop_depth: %2d", cur->loop_index(), cur->loop_depth()); | |
| 1009 } | |
| 1010 } | |
| 1011 | |
| 1012 if (TraceLinearScanLevel >= 1) { | |
| 1013 tty->print_cr("----- linear-scan block order:"); | |
| 1014 for (int block_idx = 0; block_idx < _linear_scan_order->length(); block_idx++) { | |
| 1015 BlockBegin* cur = _linear_scan_order->at(block_idx); | |
| 1016 tty->print("%4d: B%2d loop: %2d depth: %2d", cur->linear_scan_number(), cur->block_id(), cur->loop_index(), cur->loop_depth()); | |
| 1017 | |
| 1018 tty->print(cur->is_set(BlockBegin::exception_entry_flag) ? " ex" : " "); | |
| 1019 tty->print(cur->is_set(BlockBegin::critical_edge_split_flag) ? " ce" : " "); | |
| 1020 tty->print(cur->is_set(BlockBegin::linear_scan_loop_header_flag) ? " lh" : " "); | |
| 1021 tty->print(cur->is_set(BlockBegin::linear_scan_loop_end_flag) ? " le" : " "); | |
| 1022 | |
| 1023 if (cur->dominator() != NULL) { | |
| 1024 tty->print(" dom: B%d ", cur->dominator()->block_id()); | |
| 1025 } else { | |
| 1026 tty->print(" dom: NULL "); | |
| 1027 } | |
| 1028 | |
| 1029 if (cur->number_of_preds() > 0) { | |
| 1030 tty->print(" preds: "); | |
| 1031 for (int j = 0; j < cur->number_of_preds(); j++) { | |
| 1032 BlockBegin* pred = cur->pred_at(j); | |
| 1033 tty->print("B%d ", pred->block_id()); | |
| 1034 } | |
| 1035 } | |
| 1036 if (cur->number_of_sux() > 0) { | |
| 1037 tty->print(" sux: "); | |
| 1038 for (int j = 0; j < cur->number_of_sux(); j++) { | |
| 1039 BlockBegin* sux = cur->sux_at(j); | |
| 1040 tty->print("B%d ", sux->block_id()); | |
| 1041 } | |
| 1042 } | |
| 1043 if (cur->number_of_exception_handlers() > 0) { | |
| 1044 tty->print(" ex: "); | |
| 1045 for (int j = 0; j < cur->number_of_exception_handlers(); j++) { | |
| 1046 BlockBegin* ex = cur->exception_handler_at(j); | |
| 1047 tty->print("B%d ", ex->block_id()); | |
| 1048 } | |
| 1049 } | |
| 1050 tty->cr(); | |
| 1051 } | |
| 1052 } | |
| 1053 } | |
| 1054 #endif | |
| 1055 | |
| 1056 #ifdef ASSERT | |
| 1057 void ComputeLinearScanOrder::verify() { | |
| 1058 assert(_linear_scan_order->length() == _num_blocks, "wrong number of blocks in list"); | |
| 1059 | |
| 1060 if (StressLinearScan) { | |
| 1061 // blocks are scrambled when StressLinearScan is used | |
| 1062 return; | |
| 1063 } | |
| 1064 | |
| 1065 // check that all successors of a block have a higher linear-scan-number | |
| 1066 // and that all predecessors of a block have a lower linear-scan-number | |
| 1067 // (only backward branches of loops are ignored) | |
| 1068 int i; | |
| 1069 for (i = 0; i < _linear_scan_order->length(); i++) { | |
| 1070 BlockBegin* cur = _linear_scan_order->at(i); | |
| 1071 | |
| 1072 assert(cur->linear_scan_number() == i, "incorrect linear_scan_number"); | |
| 1073 assert(cur->linear_scan_number() >= 0 && cur->linear_scan_number() == _linear_scan_order->index_of(cur), "incorrect linear_scan_number"); | |
| 1074 | |
| 1075 int j; | |
| 1076 for (j = cur->number_of_sux() - 1; j >= 0; j--) { | |
| 1077 BlockBegin* sux = cur->sux_at(j); | |
| 1078 | |
| 1079 assert(sux->linear_scan_number() >= 0 && sux->linear_scan_number() == _linear_scan_order->index_of(sux), "incorrect linear_scan_number"); | |
| 1080 if (!cur->is_set(BlockBegin::linear_scan_loop_end_flag)) { | |
| 1081 assert(cur->linear_scan_number() < sux->linear_scan_number(), "invalid order"); | |
| 1082 } | |
| 1083 if (cur->loop_depth() == sux->loop_depth()) { | |
| 1084 assert(cur->loop_index() == sux->loop_index() || sux->is_set(BlockBegin::linear_scan_loop_header_flag), "successing blocks with same loop depth must have same loop index"); | |
| 1085 } | |
| 1086 } | |
| 1087 | |
| 1088 for (j = cur->number_of_preds() - 1; j >= 0; j--) { | |
| 1089 BlockBegin* pred = cur->pred_at(j); | |
| 1090 | |
| 1091 assert(pred->linear_scan_number() >= 0 && pred->linear_scan_number() == _linear_scan_order->index_of(pred), "incorrect linear_scan_number"); | |
| 1092 if (!cur->is_set(BlockBegin::linear_scan_loop_header_flag)) { | |
| 1093 assert(cur->linear_scan_number() > pred->linear_scan_number(), "invalid order"); | |
| 1094 } | |
| 1095 if (cur->loop_depth() == pred->loop_depth()) { | |
| 1096 assert(cur->loop_index() == pred->loop_index() || cur->is_set(BlockBegin::linear_scan_loop_header_flag), "successing blocks with same loop depth must have same loop index"); | |
| 1097 } | |
| 1098 | |
| 1099 assert(cur->dominator()->linear_scan_number() <= cur->pred_at(j)->linear_scan_number(), "dominator must be before predecessors"); | |
| 1100 } | |
| 1101 | |
| 1102 // check dominator | |
| 1103 if (i == 0) { | |
| 1104 assert(cur->dominator() == NULL, "first block has no dominator"); | |
| 1105 } else { | |
| 1106 assert(cur->dominator() != NULL, "all but first block must have dominator"); | |
| 1107 } | |
| 1108 assert(cur->number_of_preds() != 1 || cur->dominator() == cur->pred_at(0), "Single predecessor must also be dominator"); | |
| 1109 } | |
| 1110 | |
| 1111 // check that all loops are continuous | |
| 1112 for (int loop_idx = 0; loop_idx < _num_loops; loop_idx++) { | |
| 1113 int block_idx = 0; | |
| 1114 assert(!is_block_in_loop(loop_idx, _linear_scan_order->at(block_idx)), "the first block must not be present in any loop"); | |
| 1115 | |
| 1116 // skip blocks before the loop | |
| 1117 while (block_idx < _num_blocks && !is_block_in_loop(loop_idx, _linear_scan_order->at(block_idx))) { | |
| 1118 block_idx++; | |
| 1119 } | |
| 1120 // skip blocks of loop | |
| 1121 while (block_idx < _num_blocks && is_block_in_loop(loop_idx, _linear_scan_order->at(block_idx))) { | |
| 1122 block_idx++; | |
| 1123 } | |
| 1124 // after the first non-loop block, there must not be another loop-block | |
| 1125 while (block_idx < _num_blocks) { | |
| 1126 assert(!is_block_in_loop(loop_idx, _linear_scan_order->at(block_idx)), "loop not continuous in linear-scan order"); | |
| 1127 block_idx++; | |
| 1128 } | |
| 1129 } | |
| 1130 } | |
| 1131 #endif | |
| 1132 | |
| 1133 | |
| 1134 void IR::compute_code() { | |
| 1135 assert(is_valid(), "IR must be valid"); | |
| 1136 | |
| 1137 ComputeLinearScanOrder compute_order(start()); | |
| 1138 _num_loops = compute_order.num_loops(); | |
| 1139 _code = compute_order.linear_scan_order(); | |
| 1140 } | |
| 1141 | |
| 1142 | |
| 1143 void IR::compute_use_counts() { | |
| 1144 // make sure all values coming out of this block get evaluated. | |
| 1145 int num_blocks = _code->length(); | |
| 1146 for (int i = 0; i < num_blocks; i++) { | |
| 1147 _code->at(i)->end()->state()->pin_stack_for_linear_scan(); | |
| 1148 } | |
| 1149 | |
| 1150 // compute use counts | |
| 1151 UseCountComputer::compute(_code); | |
| 1152 } | |
| 1153 | |
| 1154 | |
| 1155 void IR::iterate_preorder(BlockClosure* closure) { | |
| 1156 assert(is_valid(), "IR must be valid"); | |
| 1157 start()->iterate_preorder(closure); | |
| 1158 } | |
| 1159 | |
| 1160 | |
| 1161 void IR::iterate_postorder(BlockClosure* closure) { | |
| 1162 assert(is_valid(), "IR must be valid"); | |
| 1163 start()->iterate_postorder(closure); | |
| 1164 } | |
| 1165 | |
| 1166 void IR::iterate_linear_scan_order(BlockClosure* closure) { | |
| 1167 linear_scan_order()->iterate_forward(closure); | |
| 1168 } | |
| 1169 | |
| 1170 | |
| 1171 #ifndef PRODUCT | |
| 1172 class BlockPrinter: public BlockClosure { | |
| 1173 private: | |
| 1174 InstructionPrinter* _ip; | |
| 1175 bool _cfg_only; | |
| 1176 bool _live_only; | |
| 1177 | |
| 1178 public: | |
| 1179 BlockPrinter(InstructionPrinter* ip, bool cfg_only, bool live_only = false) { | |
| 1180 _ip = ip; | |
| 1181 _cfg_only = cfg_only; | |
| 1182 _live_only = live_only; | |
| 1183 } | |
| 1184 | |
| 1185 virtual void block_do(BlockBegin* block) { | |
| 1186 if (_cfg_only) { | |
| 1187 _ip->print_instr(block); tty->cr(); | |
| 1188 } else { | |
| 1189 block->print_block(*_ip, _live_only); | |
| 1190 } | |
| 1191 } | |
| 1192 }; | |
| 1193 | |
| 1194 | |
| 1195 void IR::print(BlockBegin* start, bool cfg_only, bool live_only) { | |
| 1196 ttyLocker ttyl; | |
| 1197 InstructionPrinter ip(!cfg_only); | |
| 1198 BlockPrinter bp(&ip, cfg_only, live_only); | |
| 1199 start->iterate_preorder(&bp); | |
| 1200 tty->cr(); | |
| 1201 } | |
| 1202 | |
| 1203 void IR::print(bool cfg_only, bool live_only) { | |
| 1204 if (is_valid()) { | |
| 1205 print(start(), cfg_only, live_only); | |
| 1206 } else { | |
| 1207 tty->print_cr("invalid IR"); | |
| 1208 } | |
| 1209 } | |
| 1210 | |
| 1211 | |
| 1212 define_array(BlockListArray, BlockList*) | |
| 1213 define_stack(BlockListList, BlockListArray) | |
| 1214 | |
| 1215 class PredecessorValidator : public BlockClosure { | |
| 1216 private: | |
| 1217 BlockListList* _predecessors; | |
| 1218 BlockList* _blocks; | |
| 1219 | |
| 1220 static int cmp(BlockBegin** a, BlockBegin** b) { | |
| 1221 return (*a)->block_id() - (*b)->block_id(); | |
| 1222 } | |
| 1223 | |
| 1224 public: | |
| 1225 PredecessorValidator(IR* hir) { | |
| 1226 ResourceMark rm; | |
| 1227 _predecessors = new BlockListList(BlockBegin::number_of_blocks(), NULL); | |
| 1228 _blocks = new BlockList(); | |
| 1229 | |
| 1230 int i; | |
| 1231 hir->start()->iterate_preorder(this); | |
| 1232 if (hir->code() != NULL) { | |
| 1233 assert(hir->code()->length() == _blocks->length(), "must match"); | |
| 1234 for (i = 0; i < _blocks->length(); i++) { | |
| 1235 assert(hir->code()->contains(_blocks->at(i)), "should be in both lists"); | |
| 1236 } | |
| 1237 } | |
| 1238 | |
| 1239 for (i = 0; i < _blocks->length(); i++) { | |
| 1240 BlockBegin* block = _blocks->at(i); | |
| 1241 BlockList* preds = _predecessors->at(block->block_id()); | |
| 1242 if (preds == NULL) { | |
| 1243 assert(block->number_of_preds() == 0, "should be the same"); | |
| 1244 continue; | |
| 1245 } | |
| 1246 | |
| 1247 // clone the pred list so we can mutate it | |
| 1248 BlockList* pred_copy = new BlockList(); | |
| 1249 int j; | |
| 1250 for (j = 0; j < block->number_of_preds(); j++) { | |
| 1251 pred_copy->append(block->pred_at(j)); | |
| 1252 } | |
| 1253 // sort them in the same order | |
| 1254 preds->sort(cmp); | |
| 1255 pred_copy->sort(cmp); | |
| 1256 int length = MIN2(preds->length(), block->number_of_preds()); | |
| 1257 for (j = 0; j < block->number_of_preds(); j++) { | |
| 1258 assert(preds->at(j) == pred_copy->at(j), "must match"); | |
| 1259 } | |
| 1260 | |
| 1261 assert(preds->length() == block->number_of_preds(), "should be the same"); | |
| 1262 } | |
| 1263 } | |
| 1264 | |
| 1265 virtual void block_do(BlockBegin* block) { | |
| 1266 _blocks->append(block); | |
| 1267 BlockEnd* be = block->end(); | |
| 1268 int n = be->number_of_sux(); | |
| 1269 int i; | |
| 1270 for (i = 0; i < n; i++) { | |
| 1271 BlockBegin* sux = be->sux_at(i); | |
| 1272 assert(!sux->is_set(BlockBegin::exception_entry_flag), "must not be xhandler"); | |
| 1273 | |
| 1274 BlockList* preds = _predecessors->at_grow(sux->block_id(), NULL); | |
| 1275 if (preds == NULL) { | |
| 1276 preds = new BlockList(); | |
| 1277 _predecessors->at_put(sux->block_id(), preds); | |
| 1278 } | |
| 1279 preds->append(block); | |
| 1280 } | |
| 1281 | |
| 1282 n = block->number_of_exception_handlers(); | |
| 1283 for (i = 0; i < n; i++) { | |
| 1284 BlockBegin* sux = block->exception_handler_at(i); | |
| 1285 assert(sux->is_set(BlockBegin::exception_entry_flag), "must be xhandler"); | |
| 1286 | |
| 1287 BlockList* preds = _predecessors->at_grow(sux->block_id(), NULL); | |
| 1288 if (preds == NULL) { | |
| 1289 preds = new BlockList(); | |
| 1290 _predecessors->at_put(sux->block_id(), preds); | |
| 1291 } | |
| 1292 preds->append(block); | |
| 1293 } | |
| 1294 } | |
| 1295 }; | |
| 1296 | |
| 1297 void IR::verify() { | |
| 1298 #ifdef ASSERT | |
| 1299 PredecessorValidator pv(this); | |
| 1300 #endif | |
| 1301 } | |
| 1302 | |
| 1303 #endif // PRODUCT | |
| 1304 | |
| 1305 void SubstitutionResolver::substitute(Value* v) { | |
| 1306 Value v0 = *v; | |
| 1307 if (v0) { | |
| 1308 Value vs = v0->subst(); | |
| 1309 if (vs != v0) { | |
| 1310 *v = v0->subst(); | |
| 1311 } | |
| 1312 } | |
| 1313 } | |
| 1314 | |
| 1315 #ifdef ASSERT | |
| 1316 void check_substitute(Value* v) { | |
| 1317 Value v0 = *v; | |
| 1318 if (v0) { | |
| 1319 Value vs = v0->subst(); | |
| 1320 assert(vs == v0, "missed substitution"); | |
| 1321 } | |
| 1322 } | |
| 1323 #endif | |
| 1324 | |
| 1325 | |
| 1326 void SubstitutionResolver::block_do(BlockBegin* block) { | |
| 1327 Instruction* last = NULL; | |
| 1328 for (Instruction* n = block; n != NULL;) { | |
| 1329 n->values_do(substitute); | |
| 1330 // need to remove this instruction from the instruction stream | |
| 1331 if (n->subst() != n) { | |
| 1332 assert(last != NULL, "must have last"); | |
| 1333 last->set_next(n->next(), n->next()->bci()); | |
| 1334 } else { | |
| 1335 last = n; | |
| 1336 } | |
| 1337 n = last->next(); | |
| 1338 } | |
| 1339 | |
| 1340 #ifdef ASSERT | |
| 1341 if (block->state()) block->state()->values_do(check_substitute); | |
| 1342 block->block_values_do(check_substitute); | |
| 1343 if (block->end() && block->end()->state()) block->end()->state()->values_do(check_substitute); | |
| 1344 #endif | |
| 1345 } |