1 /*
   2  * Copyright (c) 1999, 2012, 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 "c1/c1_Compilation.hpp"
  27 #include "c1/c1_FrameMap.hpp"
  28 #include "c1/c1_GraphBuilder.hpp"
  29 #include "c1/c1_IR.hpp"
  30 #include "c1/c1_InstructionPrinter.hpp"
  31 #include "c1/c1_Optimizer.hpp"
  32 #include "utilities/bitMap.inline.hpp"
  33 
  34 
  35 // Implementation of XHandlers
  36 //
  37 // Note: This code could eventually go away if we are
  38 //       just using the ciExceptionHandlerStream.
  39 
  40 XHandlers::XHandlers(ciMethod* method) : _list(method->exception_table_length()) {
  41   ciExceptionHandlerStream s(method);
  42   while (!s.is_done()) {
  43     _list.append(new XHandler(s.handler()));
  44     s.next();
  45   }
  46   assert(s.count() == method->exception_table_length(), "exception table lengths inconsistent");
  47 }
  48 
  49 // deep copy of all XHandler contained in list
  50 XHandlers::XHandlers(XHandlers* other) :
  51   _list(other->length())
  52 {
  53   for (int i = 0; i < other->length(); i++) {
  54     _list.append(new XHandler(other->handler_at(i)));
  55   }
  56 }
  57 
  58 // Returns whether a particular exception type can be caught.  Also
  59 // returns true if klass is unloaded or any exception handler
  60 // classes are unloaded.  type_is_exact indicates whether the throw
  61 // is known to be exactly that class or it might throw a subtype.
  62 bool XHandlers::could_catch(ciInstanceKlass* klass, bool type_is_exact) const {
  63   // the type is unknown so be conservative
  64   if (!klass->is_loaded()) {
  65     return true;
  66   }
  67 
  68   for (int i = 0; i < length(); i++) {
  69     XHandler* handler = handler_at(i);
  70     if (handler->is_catch_all()) {
  71       // catch of ANY
  72       return true;
  73     }
  74     ciInstanceKlass* handler_klass = handler->catch_klass();
  75     // if it's unknown it might be catchable
  76     if (!handler_klass->is_loaded()) {
  77       return true;
  78     }
  79     // if the throw type is definitely a subtype of the catch type
  80     // then it can be caught.
  81     if (klass->is_subtype_of(handler_klass)) {
  82       return true;
  83     }
  84     if (!type_is_exact) {
  85       // If the type isn't exactly known then it can also be caught by
  86       // catch statements where the inexact type is a subtype of the
  87       // catch type.
  88       // given: foo extends bar extends Exception
  89       // throw bar can be caught by catch foo, catch bar, and catch
  90       // Exception, however it can't be caught by any handlers without
  91       // bar in its type hierarchy.
  92       if (handler_klass->is_subtype_of(klass)) {
  93         return true;
  94       }
  95     }
  96   }
  97 
  98   return false;
  99 }
 100 
 101 
 102 bool XHandlers::equals(XHandlers* others) const {
 103   if (others == NULL) return false;
 104   if (length() != others->length()) return false;
 105 
 106   for (int i = 0; i < length(); i++) {
 107     if (!handler_at(i)->equals(others->handler_at(i))) return false;
 108   }
 109   return true;
 110 }
 111 
 112 bool XHandler::equals(XHandler* other) const {
 113   assert(entry_pco() != -1 && other->entry_pco() != -1, "must have entry_pco");
 114 
 115   if (entry_pco() != other->entry_pco()) return false;
 116   if (scope_count() != other->scope_count()) return false;
 117   if (_desc != other->_desc) return false;
 118 
 119   assert(entry_block() == other->entry_block(), "entry_block must be equal when entry_pco is equal");
 120   return true;
 121 }
 122 
 123 
 124 // Implementation of IRScope
 125 BlockBegin* IRScope::build_graph(Compilation* compilation, int osr_bci) {
 126   GraphBuilder gm(compilation, this);
 127   NOT_PRODUCT(if (PrintValueNumbering && Verbose) gm.print_stats());
 128   if (compilation->bailed_out()) return NULL;
 129   return gm.start();
 130 }
 131 
 132 
 133 IRScope::IRScope(Compilation* compilation, IRScope* caller, int caller_bci, ciMethod* method, int osr_bci, bool create_graph)
 134 : _callees(2)
 135 , _compilation(compilation)
 136 , _requires_phi_function(method->max_locals())
 137 {
 138   _caller             = caller;
 139   _level              = caller == NULL ?  0 : caller->level() + 1;
 140   _method             = method;
 141   _xhandlers          = new XHandlers(method);
 142   _number_of_locks    = 0;
 143   _monitor_pairing_ok = method->has_balanced_monitors();
 144   _wrote_final        = false;
 145   _start              = NULL;
 146 
 147   if (osr_bci == -1) {
 148     _requires_phi_function.clear();
 149   } else {
 150         // selective creation of phi functions is not possibel in osr-methods
 151     _requires_phi_function.set_range(0, method->max_locals());
 152   }
 153 
 154   assert(method->holder()->is_loaded() , "method holder must be loaded");
 155 
 156   // build graph if monitor pairing is ok
 157   if (create_graph && monitor_pairing_ok()) _start = build_graph(compilation, osr_bci);
 158 }
 159 
 160 
 161 int IRScope::max_stack() const {
 162   int my_max = method()->max_stack();
 163   int callee_max = 0;
 164   for (int i = 0; i < number_of_callees(); i++) {
 165     callee_max = MAX2(callee_max, callee_no(i)->max_stack());
 166   }
 167   return my_max + callee_max;
 168 }
 169 
 170 
 171 bool IRScopeDebugInfo::should_reexecute() {
 172   ciMethod* cur_method = scope()->method();
 173   int       cur_bci    = bci();
 174   if (cur_method != NULL && cur_bci != SynchronizationEntryBCI) {
 175     Bytecodes::Code code = cur_method->java_code_at_bci(cur_bci);
 176     return Interpreter::bytecode_should_reexecute(code);
 177   } else
 178     return false;
 179 }
 180 
 181 
 182 // Implementation of CodeEmitInfo
 183 
 184 // Stack must be NON-null
 185 CodeEmitInfo::CodeEmitInfo(ValueStack* stack, XHandlers* exception_handlers, bool deoptimize_on_exception)
 186   : _scope(stack->scope())
 187   , _scope_debug_info(NULL)
 188   , _oop_map(NULL)
 189   , _stack(stack)
 190   , _exception_handlers(exception_handlers)
 191   , _is_method_handle_invoke(false) 
 192   , _deoptimize_on_exception(deoptimize_on_exception) {
 193   assert(_stack != NULL, "must be non null");
 194 }
 195 
 196 
 197 CodeEmitInfo::CodeEmitInfo(CodeEmitInfo* info, ValueStack* stack)
 198   : _scope(info->_scope)
 199   , _exception_handlers(NULL)
 200   , _scope_debug_info(NULL)
 201   , _oop_map(NULL)
 202   , _stack(stack == NULL ? info->_stack : stack)
 203   , _is_method_handle_invoke(info->_is_method_handle_invoke)
 204   , _deoptimize_on_exception(info->_deoptimize_on_exception) {
 205 
 206   // deep copy of exception handlers
 207   if (info->_exception_handlers != NULL) {
 208     _exception_handlers = new XHandlers(info->_exception_handlers);
 209   }
 210 }
 211 
 212 
 213 void CodeEmitInfo::record_debug_info(DebugInformationRecorder* recorder, int pc_offset) {
 214   // record the safepoint before recording the debug info for enclosing scopes
 215   recorder->add_safepoint(pc_offset, _oop_map->deep_copy());
 216   _scope_debug_info->record_debug_info(recorder, pc_offset, true/*topmost*/, _is_method_handle_invoke);
 217   recorder->end_safepoint(pc_offset);
 218 }
 219 
 220 
 221 void CodeEmitInfo::add_register_oop(LIR_Opr opr) {
 222   assert(_oop_map != NULL, "oop map must already exist");
 223   assert(opr->is_single_cpu(), "should not call otherwise");
 224 
 225   VMReg name = frame_map()->regname(opr);
 226   _oop_map->set_oop(name);
 227 }
 228 
 229 
 230 
 231 
 232 // Implementation of IR
 233 
 234 IR::IR(Compilation* compilation, ciMethod* method, int osr_bci) :
 235     _locals_size(in_WordSize(-1))
 236   , _num_loops(0) {
 237   // setup IR fields
 238   _compilation = compilation;
 239   _top_scope   = new IRScope(compilation, NULL, -1, method, osr_bci, true);
 240   _code        = NULL;
 241 }
 242 
 243 
 244 void IR::optimize_blocks() {
 245   Optimizer opt(this);
 246   if (!compilation()->profile_branches()) {
 247     if (DoCEE) {
 248       opt.eliminate_conditional_expressions();
 249 #ifndef PRODUCT
 250       if (PrintCFG || PrintCFG1) { tty->print_cr("CFG after CEE"); print(true); }
 251       if (PrintIR  || PrintIR1 ) { tty->print_cr("IR after CEE"); print(false); }
 252 #endif
 253     }
 254     if (EliminateBlocks) {
 255       opt.eliminate_blocks();
 256 #ifndef PRODUCT
 257       if (PrintCFG || PrintCFG1) { tty->print_cr("CFG after block elimination"); print(true); }
 258       if (PrintIR  || PrintIR1 ) { tty->print_cr("IR after block elimination"); print(false); }
 259 #endif
 260     }
 261   }
 262 }
 263 
 264 void IR::eliminate_null_checks() {
 265   Optimizer opt(this);
 266   if (EliminateNullChecks) {
 267     opt.eliminate_null_checks();
 268 #ifndef PRODUCT
 269     if (PrintCFG || PrintCFG1) { tty->print_cr("CFG after null check elimination"); print(true); }
 270     if (PrintIR  || PrintIR1 ) { tty->print_cr("IR after null check elimination"); print(false); }
 271 #endif
 272   }
 273 }
 274 
 275 
 276 static int sort_pairs(BlockPair** a, BlockPair** b) {
 277   if ((*a)->from() == (*b)->from()) {
 278     return (*a)->to()->block_id() - (*b)->to()->block_id();
 279   } else {
 280     return (*a)->from()->block_id() - (*b)->from()->block_id();
 281   }
 282 }
 283 
 284 
 285 class CriticalEdgeFinder: public BlockClosure {
 286   BlockPairList blocks;
 287   IR*       _ir;
 288 
 289  public:
 290   CriticalEdgeFinder(IR* ir): _ir(ir) {}
 291   void block_do(BlockBegin* bb) {
 292     BlockEnd* be = bb->end();
 293     int nos = be->number_of_sux();
 294     if (nos >= 2) {
 295       for (int i = 0; i < nos; i++) {
 296         BlockBegin* sux = be->sux_at(i);
 297         if (sux->number_of_preds() >= 2) {
 298           blocks.append(new BlockPair(bb, sux));
 299         }
 300       }
 301     }
 302   }
 303 
 304   void split_edges() {
 305     BlockPair* last_pair = NULL;
 306     blocks.sort(sort_pairs);
 307     for (int i = 0; i < blocks.length(); i++) {
 308       BlockPair* pair = blocks.at(i);
 309       if (last_pair != NULL && pair->is_same(last_pair)) continue;
 310       BlockBegin* from = pair->from();
 311       BlockBegin* to = pair->to();
 312       BlockBegin* split = from->insert_block_between(to);
 313 #ifndef PRODUCT
 314       if ((PrintIR || PrintIR1) && Verbose) {
 315         tty->print_cr("Split critical edge B%d -> B%d (new block B%d)",
 316                       from->block_id(), to->block_id(), split->block_id());
 317       }
 318 #endif
 319       last_pair = pair;
 320     }
 321   }
 322 };
 323 
 324 void IR::split_critical_edges() {
 325   CriticalEdgeFinder cef(this);
 326 
 327   iterate_preorder(&cef);
 328   cef.split_edges();
 329 }
 330 
 331 
 332 class UseCountComputer: public ValueVisitor, BlockClosure {
 333  private:
 334   void visit(Value* n) {
 335     // Local instructions and Phis for expression stack values at the
 336     // start of basic blocks are not added to the instruction list
 337     if (!(*n)->is_linked() && (*n)->can_be_linked()) {
 338       assert(false, "a node was not appended to the graph");
 339       Compilation::current()->bailout("a node was not appended to the graph");
 340     }
 341     // use n's input if not visited before
 342     if (!(*n)->is_pinned() && !(*n)->has_uses()) {
 343       // note: a) if the instruction is pinned, it will be handled by compute_use_count
 344       //       b) if the instruction has uses, it was touched before
 345       //       => in both cases we don't need to update n's values
 346       uses_do(n);
 347     }
 348     // use n
 349     (*n)->_use_count++;
 350   }
 351 
 352   Values* worklist;
 353   int depth;
 354   enum {
 355     max_recurse_depth = 20
 356   };
 357 
 358   void uses_do(Value* n) {
 359     depth++;
 360     if (depth > max_recurse_depth) {
 361       // don't allow the traversal to recurse too deeply
 362       worklist->push(*n);
 363     } else {
 364       (*n)->input_values_do(this);
 365       // special handling for some instructions
 366       if ((*n)->as_BlockEnd() != NULL) {
 367         // note on BlockEnd:
 368         //   must 'use' the stack only if the method doesn't
 369         //   terminate, however, in those cases stack is empty
 370         (*n)->state_values_do(this);
 371       }
 372     }
 373     depth--;
 374   }
 375 
 376   void block_do(BlockBegin* b) {
 377     depth = 0;
 378     // process all pinned nodes as the roots of expression trees
 379     for (Instruction* n = b; n != NULL; n = n->next()) {
 380       if (n->is_pinned()) uses_do(&n);
 381     }
 382     assert(depth == 0, "should have counted back down");
 383 
 384     // now process any unpinned nodes which recursed too deeply
 385     while (worklist->length() > 0) {
 386       Value t = worklist->pop();
 387       if (!t->is_pinned()) {
 388         // compute the use count
 389         uses_do(&t);
 390 
 391         // pin the instruction so that LIRGenerator doesn't recurse
 392         // too deeply during it's evaluation.
 393         t->pin();
 394       }
 395     }
 396     assert(depth == 0, "should have counted back down");
 397   }
 398 
 399   UseCountComputer() {
 400     worklist = new Values();
 401     depth = 0;
 402   }
 403 
 404  public:
 405   static void compute(BlockList* blocks) {
 406     UseCountComputer ucc;
 407     blocks->iterate_backward(&ucc);
 408   }
 409 };
 410 
 411 
 412 // helper macro for short definition of trace-output inside code
 413 #ifndef PRODUCT
 414   #define TRACE_LINEAR_SCAN(level, code)       \
 415     if (TraceLinearScanLevel >= level) {       \
 416       code;                                    \
 417     }
 418 #else
 419   #define TRACE_LINEAR_SCAN(level, code)
 420 #endif
 421 
 422 class ComputeLinearScanOrder : public StackObj {
 423  private:
 424   int        _max_block_id;        // the highest block_id of a block
 425   int        _num_blocks;          // total number of blocks (smaller than _max_block_id)
 426   int        _num_loops;           // total number of loops
 427   bool       _iterative_dominators;// method requires iterative computation of dominatiors
 428 
 429   BlockList* _linear_scan_order;   // the resulting list of blocks in correct order
 430 
 431   BitMap     _visited_blocks;      // used for recursive processing of blocks
 432   BitMap     _active_blocks;       // used for recursive processing of blocks
 433   BitMap     _dominator_blocks;    // temproary BitMap used for computation of dominator
 434   intArray   _forward_branches;    // number of incoming forward branches for each block
 435   BlockList  _loop_end_blocks;     // list of all loop end blocks collected during count_edges
 436   BitMap2D   _loop_map;            // two-dimensional bit set: a bit is set if a block is contained in a loop
 437   BlockList  _work_list;           // temporary list (used in mark_loops and compute_order)
 438   BlockList  _loop_headers;
 439 
 440   Compilation* _compilation;
 441 
 442   // accessors for _visited_blocks and _active_blocks
 443   void init_visited()                     { _active_blocks.clear(); _visited_blocks.clear(); }
 444   bool is_visited(BlockBegin* b) const    { return _visited_blocks.at(b->block_id()); }
 445   bool is_active(BlockBegin* b) const     { return _active_blocks.at(b->block_id()); }
 446   void set_visited(BlockBegin* b)         { assert(!is_visited(b), "already set"); _visited_blocks.set_bit(b->block_id()); }
 447   void set_active(BlockBegin* b)          { assert(!is_active(b), "already set");  _active_blocks.set_bit(b->block_id()); }
 448   void clear_active(BlockBegin* b)        { assert(is_active(b), "not already");   _active_blocks.clear_bit(b->block_id()); }
 449 
 450   // accessors for _forward_branches
 451   void inc_forward_branches(BlockBegin* b) { _forward_branches.at_put(b->block_id(), _forward_branches.at(b->block_id()) + 1); }
 452   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()); }
 453 
 454   // accessors for _loop_map
 455   bool is_block_in_loop   (int loop_idx, BlockBegin* b) const { return _loop_map.at(loop_idx, b->block_id()); }
 456   void set_block_in_loop  (int loop_idx, BlockBegin* b)       { _loop_map.set_bit(loop_idx, b->block_id()); }
 457   void clear_block_in_loop(int loop_idx, int block_id)        { _loop_map.clear_bit(loop_idx, block_id); }
 458 
 459   // count edges between blocks
 460   void count_edges(BlockBegin* cur, BlockBegin* parent);
 461 
 462   // loop detection
 463   void mark_loops();
 464   void clear_non_natural_loops(BlockBegin* start_block);
 465   void assign_loop_depth(BlockBegin* start_block);
 466 
 467   // computation of final block order
 468   BlockBegin* common_dominator(BlockBegin* a, BlockBegin* b);
 469   void compute_dominator(BlockBegin* cur, BlockBegin* parent);
 470   int  compute_weight(BlockBegin* cur);
 471   bool ready_for_processing(BlockBegin* cur);
 472   void sort_into_work_list(BlockBegin* b);
 473   void append_block(BlockBegin* cur);
 474   void compute_order(BlockBegin* start_block);
 475 
 476   // fixup of dominators for non-natural loops
 477   bool compute_dominators_iter();
 478   void compute_dominators();
 479 
 480   // debug functions
 481   NOT_PRODUCT(void print_blocks();)
 482   DEBUG_ONLY(void verify();)
 483 
 484   Compilation* compilation() const { return _compilation; }
 485  public:
 486   ComputeLinearScanOrder(Compilation* c, BlockBegin* start_block);
 487 
 488   // accessors for final result
 489   BlockList* linear_scan_order() const    { return _linear_scan_order; }
 490   int        num_loops() const            { return _num_loops; }
 491 };
 492 
 493 
 494 ComputeLinearScanOrder::ComputeLinearScanOrder(Compilation* c, BlockBegin* start_block) :
 495   _max_block_id(BlockBegin::number_of_blocks()),
 496   _num_blocks(0),
 497   _num_loops(0),
 498   _iterative_dominators(false),
 499   _visited_blocks(_max_block_id),
 500   _active_blocks(_max_block_id),
 501   _dominator_blocks(_max_block_id),
 502   _forward_branches(_max_block_id, 0),
 503   _loop_end_blocks(8),
 504   _work_list(8),
 505   _linear_scan_order(NULL), // initialized later with correct size
 506   _loop_map(0, 0),          // initialized later with correct size
 507   _compilation(c)
 508 {
 509   TRACE_LINEAR_SCAN(2, "***** computing linear-scan block order");
 510 
 511   init_visited();
 512   count_edges(start_block, NULL);
 513 
 514   if (compilation()->is_profiling()) {
 515     ciMethod *method = compilation()->method();
 516     if (!method->is_accessor()) {
 517       ciMethodData* md = method->method_data_or_null();
 518       assert(md != NULL, "Sanity");
 519       md->set_compilation_stats(_num_loops, _num_blocks);
 520     }
 521   }
 522 
 523   if (_num_loops > 0) {
 524     mark_loops();
 525     clear_non_natural_loops(start_block);
 526     assign_loop_depth(start_block);
 527   }
 528 
 529   compute_order(start_block);
 530   compute_dominators();
 531 
 532   NOT_PRODUCT(print_blocks());
 533   DEBUG_ONLY(verify());
 534 }
 535 
 536 
 537 // Traverse the CFG:
 538 // * count total number of blocks
 539 // * count all incoming edges and backward incoming edges
 540 // * number loop header blocks
 541 // * create a list with all loop end blocks
 542 void ComputeLinearScanOrder::count_edges(BlockBegin* cur, BlockBegin* parent) {
 543   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));
 544   assert(cur->dominator() == NULL, "dominator already initialized");
 545 
 546   if (is_active(cur)) {
 547     TRACE_LINEAR_SCAN(3, tty->print_cr("backward branch"));
 548     assert(is_visited(cur), "block must be visisted when block is active");
 549     assert(parent != NULL, "must have parent");
 550 
 551     cur->set(BlockBegin::linear_scan_loop_header_flag);
 552     cur->set(BlockBegin::backward_branch_target_flag);
 553 
 554     parent->set(BlockBegin::linear_scan_loop_end_flag);
 555 
 556     // When a loop header is also the start of an exception handler, then the backward branch is
 557     // an exception edge. Because such edges are usually critical edges which cannot be split, the
 558     // loop must be excluded here from processing.
 559     if (cur->is_set(BlockBegin::exception_entry_flag)) {
 560       // Make sure that dominators are correct in this weird situation
 561       _iterative_dominators = true;
 562       return;
 563     }
 564     assert(parent->number_of_sux() == 1 && parent->sux_at(0) == cur,
 565            "loop end blocks must have one successor (critical edges are split)");
 566 
 567     _loop_end_blocks.append(parent);
 568     return;
 569   }
 570 
 571   // increment number of incoming forward branches
 572   inc_forward_branches(cur);
 573 
 574   if (is_visited(cur)) {
 575     TRACE_LINEAR_SCAN(3, tty->print_cr("block already visited"));
 576     return;
 577   }
 578 
 579   _num_blocks++;
 580   set_visited(cur);
 581   set_active(cur);
 582 
 583   // recursive call for all successors
 584   int i;
 585   for (i = cur->number_of_sux() - 1; i >= 0; i--) {
 586     count_edges(cur->sux_at(i), cur);
 587   }
 588   for (i = cur->number_of_exception_handlers() - 1; i >= 0; i--) {
 589     count_edges(cur->exception_handler_at(i), cur);
 590   }
 591 
 592   clear_active(cur);
 593 
 594   // Each loop has a unique number.
 595   // When multiple loops are nested, assign_loop_depth assumes that the
 596   // innermost loop has the lowest number. This is guaranteed by setting
 597   // the loop number after the recursive calls for the successors above
 598   // have returned.
 599   if (cur->is_set(BlockBegin::linear_scan_loop_header_flag)) {
 600     assert(cur->loop_index() == -1, "cannot set loop-index twice");
 601     TRACE_LINEAR_SCAN(3, tty->print_cr("Block B%d is loop header of loop %d", cur->block_id(), _num_loops));
 602 
 603     cur->set_loop_index(_num_loops);
 604     _loop_headers.append(cur);
 605     _num_loops++;
 606   }
 607 
 608   TRACE_LINEAR_SCAN(3, tty->print_cr("Finished count_edges for block B%d", cur->block_id()));
 609 }
 610 
 611 
 612 void ComputeLinearScanOrder::mark_loops() {
 613   TRACE_LINEAR_SCAN(3, tty->print_cr("----- marking loops"));
 614 
 615   _loop_map = BitMap2D(_num_loops, _max_block_id);
 616   _loop_map.clear();
 617 
 618   for (int i = _loop_end_blocks.length() - 1; i >= 0; i--) {
 619     BlockBegin* loop_end   = _loop_end_blocks.at(i);
 620     BlockBegin* loop_start = loop_end->sux_at(0);
 621     int         loop_idx   = loop_start->loop_index();
 622 
 623     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));
 624     assert(loop_end->is_set(BlockBegin::linear_scan_loop_end_flag), "loop end flag must be set");
 625     assert(loop_end->number_of_sux() == 1, "incorrect number of successors");
 626     assert(loop_start->is_set(BlockBegin::linear_scan_loop_header_flag), "loop header flag must be set");
 627     assert(loop_idx >= 0 && loop_idx < _num_loops, "loop index not set");
 628     assert(_work_list.is_empty(), "work list must be empty before processing");
 629 
 630     // add the end-block of the loop to the working list
 631     _work_list.push(loop_end);
 632     set_block_in_loop(loop_idx, loop_end);
 633     do {
 634       BlockBegin* cur = _work_list.pop();
 635 
 636       TRACE_LINEAR_SCAN(3, tty->print_cr("    processing B%d", cur->block_id()));
 637       assert(is_block_in_loop(loop_idx, cur), "bit in loop map must be set when block is in work list");
 638 
 639       // recursive processing of all predecessors ends when start block of loop is reached
 640       if (cur != loop_start && !cur->is_set(BlockBegin::osr_entry_flag)) {
 641         for (int j = cur->number_of_preds() - 1; j >= 0; j--) {
 642           BlockBegin* pred = cur->pred_at(j);
 643 
 644           if (!is_block_in_loop(loop_idx, pred) /*&& !pred->is_set(BlockBeginosr_entry_flag)*/) {
 645             // this predecessor has not been processed yet, so add it to work list
 646             TRACE_LINEAR_SCAN(3, tty->print_cr("    pushing B%d", pred->block_id()));
 647             _work_list.push(pred);
 648             set_block_in_loop(loop_idx, pred);
 649           }
 650         }
 651       }
 652     } while (!_work_list.is_empty());
 653   }
 654 }
 655 
 656 
 657 // check for non-natural loops (loops where the loop header does not dominate
 658 // all other loop blocks = loops with mulitple entries).
 659 // such loops are ignored
 660 void ComputeLinearScanOrder::clear_non_natural_loops(BlockBegin* start_block) {
 661   for (int i = _num_loops - 1; i >= 0; i--) {
 662     if (is_block_in_loop(i, start_block)) {
 663       // loop i contains the entry block of the method
 664       // -> this is not a natural loop, so ignore it
 665       TRACE_LINEAR_SCAN(2, tty->print_cr("Loop %d is non-natural, so it is ignored", i));
 666 
 667       BlockBegin *loop_header = _loop_headers.at(i);
 668       assert(loop_header->is_set(BlockBegin::linear_scan_loop_header_flag), "Must be loop header");
 669 
 670       for (int j = 0; j < loop_header->number_of_preds(); j++)
 671       {
 672         BlockBegin *pred = loop_header->pred_at(j);
 673         pred->clear(BlockBegin::linear_scan_loop_end_flag);
 674       }
 675 
 676       loop_header->clear(BlockBegin::linear_scan_loop_header_flag);
 677 
 678       for (int block_id = _max_block_id - 1; block_id >= 0; block_id--) {
 679         clear_block_in_loop(i, block_id);
 680       }
 681       _iterative_dominators = true;
 682     }
 683   }
 684 }
 685 
 686 void ComputeLinearScanOrder::assign_loop_depth(BlockBegin* start_block) {
 687   TRACE_LINEAR_SCAN(3, "----- computing loop-depth and weight");
 688   init_visited();
 689 
 690   assert(_work_list.is_empty(), "work list must be empty before processing");
 691   _work_list.append(start_block);
 692 
 693   do {
 694     BlockBegin* cur = _work_list.pop();
 695 
 696     if (!is_visited(cur)) {
 697       set_visited(cur);
 698       TRACE_LINEAR_SCAN(4, tty->print_cr("Computing loop depth for block B%d", cur->block_id()));
 699 
 700       // compute loop-depth and loop-index for the block
 701       assert(cur->loop_depth() == 0, "cannot set loop-depth twice");
 702       int i;
 703       int loop_depth = 0;
 704       int min_loop_idx = -1;
 705       for (i = _num_loops - 1; i >= 0; i--) {
 706         if (is_block_in_loop(i, cur)) {
 707           loop_depth++;
 708           min_loop_idx = i;
 709         }
 710       }
 711       cur->set_loop_depth(loop_depth);
 712       cur->set_loop_index(min_loop_idx);
 713 
 714       // append all unvisited successors to work list
 715       for (i = cur->number_of_sux() - 1; i >= 0; i--) {
 716         _work_list.append(cur->sux_at(i));
 717       }
 718       for (i = cur->number_of_exception_handlers() - 1; i >= 0; i--) {
 719         _work_list.append(cur->exception_handler_at(i));
 720       }
 721     }
 722   } while (!_work_list.is_empty());
 723 }
 724 
 725 
 726 BlockBegin* ComputeLinearScanOrder::common_dominator(BlockBegin* a, BlockBegin* b) {
 727   assert(a != NULL && b != NULL, "must have input blocks");
 728 
 729   _dominator_blocks.clear();
 730   while (a != NULL) {
 731     _dominator_blocks.set_bit(a->block_id());
 732     assert(a->dominator() != NULL || a == _linear_scan_order->at(0), "dominator must be initialized");
 733     a = a->dominator();
 734   }
 735   while (b != NULL && !_dominator_blocks.at(b->block_id())) {
 736     assert(b->dominator() != NULL || b == _linear_scan_order->at(0), "dominator must be initialized");
 737     b = b->dominator();
 738   }
 739 
 740   assert(b != NULL, "could not find dominator");
 741   return b;
 742 }
 743 
 744 void ComputeLinearScanOrder::compute_dominator(BlockBegin* cur, BlockBegin* parent) {
 745   if (cur->dominator() == NULL) {
 746     TRACE_LINEAR_SCAN(4, tty->print_cr("DOM: initializing dominator of B%d to B%d", cur->block_id(), parent->block_id()));
 747     cur->set_dominator(parent);
 748 
 749   } else if (!(cur->is_set(BlockBegin::linear_scan_loop_header_flag) && parent->is_set(BlockBegin::linear_scan_loop_end_flag))) {
 750     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()));
 751     // Does not hold for exception blocks
 752     assert(cur->number_of_preds() > 1 || cur->is_set(BlockBegin::exception_entry_flag), "");
 753     cur->set_dominator(common_dominator(cur->dominator(), parent));
 754   }
 755 
 756   // Additional edge to xhandler of all our successors
 757   // range check elimination needs that the state at the end of a
 758   // block be valid in every block it dominates so cur must dominate
 759   // the exception handlers of its successors.
 760   int num_cur_xhandler = cur->number_of_exception_handlers();
 761   for (int j = 0; j < num_cur_xhandler; j++) {
 762     BlockBegin* xhandler = cur->exception_handler_at(j);
 763     compute_dominator(xhandler, parent);
 764   }
 765 }
 766 
 767 
 768 int ComputeLinearScanOrder::compute_weight(BlockBegin* cur) {
 769   BlockBegin* single_sux = NULL;
 770   if (cur->number_of_sux() == 1) {
 771     single_sux = cur->sux_at(0);
 772   }
 773 
 774   // limit loop-depth to 15 bit (only for security reason, it will never be so big)
 775   int weight = (cur->loop_depth() & 0x7FFF) << 16;
 776 
 777   // general macro for short definition of weight flags
 778   // the first instance of INC_WEIGHT_IF has the highest priority
 779   int cur_bit = 15;
 780   #define INC_WEIGHT_IF(condition) if ((condition)) { weight |= (1 << cur_bit); } cur_bit--;
 781 
 782   // this is necessery for the (very rare) case that two successing blocks have
 783   // the same loop depth, but a different loop index (can happen for endless loops
 784   // with exception handlers)
 785   INC_WEIGHT_IF(!cur->is_set(BlockBegin::linear_scan_loop_header_flag));
 786 
 787   // loop end blocks (blocks that end with a backward branch) are added
 788   // after all other blocks of the loop.
 789   INC_WEIGHT_IF(!cur->is_set(BlockBegin::linear_scan_loop_end_flag));
 790 
 791   // critical edge split blocks are prefered because than they have a bigger
 792   // proability to be completely empty
 793   INC_WEIGHT_IF(cur->is_set(BlockBegin::critical_edge_split_flag));
 794 
 795   // exceptions should not be thrown in normal control flow, so these blocks
 796   // are added as late as possible
 797   INC_WEIGHT_IF(cur->end()->as_Throw() == NULL  && (single_sux == NULL || single_sux->end()->as_Throw()  == NULL));
 798   INC_WEIGHT_IF(cur->end()->as_Return() == NULL && (single_sux == NULL || single_sux->end()->as_Return() == NULL));
 799 
 800   // exceptions handlers are added as late as possible
 801   INC_WEIGHT_IF(!cur->is_set(BlockBegin::exception_entry_flag));
 802 
 803   // guarantee that weight is > 0
 804   weight |= 1;
 805 
 806   #undef INC_WEIGHT_IF
 807   assert(cur_bit >= 0, "too many flags");
 808   assert(weight > 0, "weight cannot become negative");
 809 
 810   return weight;
 811 }
 812 
 813 bool ComputeLinearScanOrder::ready_for_processing(BlockBegin* cur) {
 814   // Discount the edge just traveled.
 815   // When the number drops to zero, all forward branches were processed
 816   if (dec_forward_branches(cur) != 0) {
 817     return false;
 818   }
 819 
 820   assert(_linear_scan_order->index_of(cur) == -1, "block already processed (block can be ready only once)");
 821   assert(_work_list.index_of(cur) == -1, "block already in work-list (block can be ready only once)");
 822   return true;
 823 }
 824 
 825 void ComputeLinearScanOrder::sort_into_work_list(BlockBegin* cur) {
 826   assert(_work_list.index_of(cur) == -1, "block already in work list");
 827 
 828   int cur_weight = compute_weight(cur);
 829 
 830   // the linear_scan_number is used to cache the weight of a block
 831   cur->set_linear_scan_number(cur_weight);
 832 
 833 #ifndef PRODUCT
 834   if (StressLinearScan) {
 835     _work_list.insert_before(0, cur);
 836     return;
 837   }
 838 #endif
 839 
 840   _work_list.append(NULL); // provide space for new element
 841 
 842   int insert_idx = _work_list.length() - 1;
 843   while (insert_idx > 0 && _work_list.at(insert_idx - 1)->linear_scan_number() > cur_weight) {
 844     _work_list.at_put(insert_idx, _work_list.at(insert_idx - 1));
 845     insert_idx--;
 846   }
 847   _work_list.at_put(insert_idx, cur);
 848 
 849   TRACE_LINEAR_SCAN(3, tty->print_cr("Sorted B%d into worklist. new worklist:", cur->block_id()));
 850   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()));
 851 
 852 #ifdef ASSERT
 853   for (int i = 0; i < _work_list.length(); i++) {
 854     assert(_work_list.at(i)->linear_scan_number() > 0, "weight not set");
 855     assert(i == 0 || _work_list.at(i - 1)->linear_scan_number() <= _work_list.at(i)->linear_scan_number(), "incorrect order in worklist");
 856   }
 857 #endif
 858 }
 859 
 860 void ComputeLinearScanOrder::append_block(BlockBegin* cur) {
 861   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()));
 862   assert(_linear_scan_order->index_of(cur) == -1, "cannot add the same block twice");
 863 
 864   // currently, the linear scan order and code emit order are equal.
 865   // therefore the linear_scan_number and the weight of a block must also
 866   // be equal.
 867   cur->set_linear_scan_number(_linear_scan_order->length());
 868   _linear_scan_order->append(cur);
 869 }
 870 
 871 void ComputeLinearScanOrder::compute_order(BlockBegin* start_block) {
 872   TRACE_LINEAR_SCAN(3, "----- computing final block order");
 873 
 874   // the start block is always the first block in the linear scan order
 875   _linear_scan_order = new BlockList(_num_blocks);
 876   append_block(start_block);
 877 
 878   assert(start_block->end()->as_Base() != NULL, "start block must end with Base-instruction");
 879   BlockBegin* std_entry = ((Base*)start_block->end())->std_entry();
 880   BlockBegin* osr_entry = ((Base*)start_block->end())->osr_entry();
 881 
 882   BlockBegin* sux_of_osr_entry = NULL;
 883   if (osr_entry != NULL) {
 884     // special handling for osr entry:
 885     // ignore the edge between the osr entry and its successor for processing
 886     // the osr entry block is added manually below
 887     assert(osr_entry->number_of_sux() == 1, "osr entry must have exactly one successor");
 888     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");
 889 
 890     sux_of_osr_entry = osr_entry->sux_at(0);
 891     dec_forward_branches(sux_of_osr_entry);
 892 
 893     compute_dominator(osr_entry, start_block);
 894     _iterative_dominators = true;
 895   }
 896   compute_dominator(std_entry, start_block);
 897 
 898   // start processing with standard entry block
 899   assert(_work_list.is_empty(), "list must be empty before processing");
 900 
 901   if (ready_for_processing(std_entry)) {
 902     sort_into_work_list(std_entry);
 903   } else {
 904     assert(false, "the std_entry must be ready for processing (otherwise, the method has no start block)");
 905   }
 906 
 907   do {
 908     BlockBegin* cur = _work_list.pop();
 909 
 910     if (cur == sux_of_osr_entry) {
 911       // the osr entry block is ignored in normal processing, it is never added to the
 912       // work list. Instead, it is added as late as possible manually here.
 913       append_block(osr_entry);
 914       compute_dominator(cur, osr_entry);
 915     }
 916     append_block(cur);
 917 
 918     int i;
 919     int num_sux = cur->number_of_sux();
 920     // changed loop order to get "intuitive" order of if- and else-blocks
 921     for (i = 0; i < num_sux; i++) {
 922       BlockBegin* sux = cur->sux_at(i);
 923       compute_dominator(sux, cur);
 924       if (ready_for_processing(sux)) {
 925         sort_into_work_list(sux);
 926       }
 927     }
 928     num_sux = cur->number_of_exception_handlers();
 929     for (i = 0; i < num_sux; i++) {
 930       BlockBegin* sux = cur->exception_handler_at(i);
 931       if (ready_for_processing(sux)) {
 932         sort_into_work_list(sux);
 933       }
 934     }
 935   } while (_work_list.length() > 0);
 936 }
 937 
 938 
 939 bool ComputeLinearScanOrder::compute_dominators_iter() {
 940   bool changed = false;
 941   int num_blocks = _linear_scan_order->length();
 942 
 943   assert(_linear_scan_order->at(0)->dominator() == NULL, "must not have dominator");
 944   assert(_linear_scan_order->at(0)->number_of_preds() == 0, "must not have predecessors");
 945   for (int i = 1; i < num_blocks; i++) {
 946     BlockBegin* block = _linear_scan_order->at(i);
 947 
 948     BlockBegin* dominator = block->pred_at(0);
 949     int num_preds = block->number_of_preds();
 950 
 951     TRACE_LINEAR_SCAN(4, tty->print_cr("DOM: Processing B%d", block->block_id()));
 952 
 953     for (int j = 0; j < num_preds; j++) {
 954 
 955       BlockBegin *pred = block->pred_at(j);
 956       TRACE_LINEAR_SCAN(4, tty->print_cr("   DOM: Subrocessing B%d", pred->block_id()));
 957 
 958       if (block->is_set(BlockBegin::exception_entry_flag)) {
 959         dominator = common_dominator(dominator, pred);
 960         int num_pred_preds = pred->number_of_preds();
 961         for (int k = 0; k < num_pred_preds; k++) {
 962           dominator = common_dominator(dominator, pred->pred_at(k));
 963         }
 964       } else {
 965         dominator = common_dominator(dominator, pred);
 966       }
 967     }
 968 
 969     if (dominator != block->dominator()) {
 970       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()));
 971 
 972       block->set_dominator(dominator);
 973       changed = true;
 974     }
 975   }
 976   return changed;
 977 }
 978 
 979 void ComputeLinearScanOrder::compute_dominators() {
 980   TRACE_LINEAR_SCAN(3, tty->print_cr("----- computing dominators (iterative computation reqired: %d)", _iterative_dominators));
 981 
 982   // iterative computation of dominators is only required for methods with non-natural loops
 983   // and OSR-methods. For all other methods, the dominators computed when generating the
 984   // linear scan block order are correct.
 985   if (_iterative_dominators) {
 986     do {
 987       TRACE_LINEAR_SCAN(1, tty->print_cr("DOM: next iteration of fix-point calculation"));
 988     } while (compute_dominators_iter());
 989   }
 990 
 991   // check that dominators are correct
 992   assert(!compute_dominators_iter(), "fix point not reached");
 993 
 994   // Add Blocks to dominates-Array
 995   int num_blocks = _linear_scan_order->length();
 996   for (int i = 0; i < num_blocks; i++) {
 997     BlockBegin* block = _linear_scan_order->at(i);
 998 
 999     BlockBegin *dom = block->dominator();
1000     if (dom) {
1001       assert(dom->dominator_depth() != -1, "Dominator must have been visited before");
1002       dom->dominates()->append(block);
1003       block->set_dominator_depth(dom->dominator_depth() + 1);
1004     } else {
1005       block->set_dominator_depth(0);
1006     }
1007   }
1008 }
1009 
1010 
1011 #ifndef PRODUCT
1012 void ComputeLinearScanOrder::print_blocks() {
1013   if (TraceLinearScanLevel >= 2) {
1014     tty->print_cr("----- loop information:");
1015     for (int block_idx = 0; block_idx < _linear_scan_order->length(); block_idx++) {
1016       BlockBegin* cur = _linear_scan_order->at(block_idx);
1017 
1018       tty->print("%4d: B%2d: ", cur->linear_scan_number(), cur->block_id());
1019       for (int loop_idx = 0; loop_idx < _num_loops; loop_idx++) {
1020         tty->print ("%d ", is_block_in_loop(loop_idx, cur));
1021       }
1022       tty->print_cr(" -> loop_index: %2d, loop_depth: %2d", cur->loop_index(), cur->loop_depth());
1023     }
1024   }
1025 
1026   if (TraceLinearScanLevel >= 1) {
1027     tty->print_cr("----- linear-scan block order:");
1028     for (int block_idx = 0; block_idx < _linear_scan_order->length(); block_idx++) {
1029       BlockBegin* cur = _linear_scan_order->at(block_idx);
1030       tty->print("%4d: B%2d    loop: %2d  depth: %2d", cur->linear_scan_number(), cur->block_id(), cur->loop_index(), cur->loop_depth());
1031 
1032       tty->print(cur->is_set(BlockBegin::exception_entry_flag)         ? " ex" : "   ");
1033       tty->print(cur->is_set(BlockBegin::critical_edge_split_flag)     ? " ce" : "   ");
1034       tty->print(cur->is_set(BlockBegin::linear_scan_loop_header_flag) ? " lh" : "   ");
1035       tty->print(cur->is_set(BlockBegin::linear_scan_loop_end_flag)    ? " le" : "   ");
1036 
1037       if (cur->dominator() != NULL) {
1038         tty->print("    dom: B%d ", cur->dominator()->block_id());
1039       } else {
1040         tty->print("    dom: NULL ");
1041       }
1042 
1043       if (cur->number_of_preds() > 0) {
1044         tty->print("    preds: ");
1045         for (int j = 0; j < cur->number_of_preds(); j++) {
1046           BlockBegin* pred = cur->pred_at(j);
1047           tty->print("B%d ", pred->block_id());
1048         }
1049       }
1050       if (cur->number_of_sux() > 0) {
1051         tty->print("    sux: ");
1052         for (int j = 0; j < cur->number_of_sux(); j++) {
1053           BlockBegin* sux = cur->sux_at(j);
1054           tty->print("B%d ", sux->block_id());
1055         }
1056       }
1057       if (cur->number_of_exception_handlers() > 0) {
1058         tty->print("    ex: ");
1059         for (int j = 0; j < cur->number_of_exception_handlers(); j++) {
1060           BlockBegin* ex = cur->exception_handler_at(j);
1061           tty->print("B%d ", ex->block_id());
1062         }
1063       }
1064       tty->cr();
1065     }
1066   }
1067 }
1068 #endif
1069 
1070 #ifdef ASSERT
1071 void ComputeLinearScanOrder::verify() {
1072   assert(_linear_scan_order->length() == _num_blocks, "wrong number of blocks in list");
1073 
1074   if (StressLinearScan) {
1075     // blocks are scrambled when StressLinearScan is used
1076     return;
1077   }
1078 
1079   // check that all successors of a block have a higher linear-scan-number
1080   // and that all predecessors of a block have a lower linear-scan-number
1081   // (only backward branches of loops are ignored)
1082   int i;
1083   for (i = 0; i < _linear_scan_order->length(); i++) {
1084     BlockBegin* cur = _linear_scan_order->at(i);
1085 
1086     assert(cur->linear_scan_number() == i, "incorrect linear_scan_number");
1087     assert(cur->linear_scan_number() >= 0 && cur->linear_scan_number() == _linear_scan_order->index_of(cur), "incorrect linear_scan_number");
1088 
1089     int j;
1090     for (j = cur->number_of_sux() - 1; j >= 0; j--) {
1091       BlockBegin* sux = cur->sux_at(j);
1092 
1093       assert(sux->linear_scan_number() >= 0 && sux->linear_scan_number() == _linear_scan_order->index_of(sux), "incorrect linear_scan_number");
1094       if (!sux->is_set(BlockBegin::backward_branch_target_flag)) {
1095         assert(cur->linear_scan_number() < sux->linear_scan_number(), "invalid order");
1096       }
1097       if (cur->loop_depth() == sux->loop_depth()) {
1098         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");
1099       }
1100     }
1101 
1102     for (j = cur->number_of_preds() - 1; j >= 0; j--) {
1103       BlockBegin* pred = cur->pred_at(j);
1104 
1105       assert(pred->linear_scan_number() >= 0 && pred->linear_scan_number() == _linear_scan_order->index_of(pred), "incorrect linear_scan_number");
1106       if (!cur->is_set(BlockBegin::backward_branch_target_flag)) {
1107         assert(cur->linear_scan_number() > pred->linear_scan_number(), "invalid order");
1108       }
1109       if (cur->loop_depth() == pred->loop_depth()) {
1110         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");
1111       }
1112 
1113       assert(cur->dominator()->linear_scan_number() <= cur->pred_at(j)->linear_scan_number(), "dominator must be before predecessors");
1114     }
1115 
1116     // check dominator
1117     if (i == 0) {
1118       assert(cur->dominator() == NULL, "first block has no dominator");
1119     } else {
1120       assert(cur->dominator() != NULL, "all but first block must have dominator");
1121     }
1122     // Assertion does not hold for exception handlers
1123     assert(cur->number_of_preds() != 1 || cur->dominator() == cur->pred_at(0) || cur->is_set(BlockBegin::exception_entry_flag), "Single predecessor must also be dominator");
1124   }
1125 
1126   // check that all loops are continuous
1127   for (int loop_idx = 0; loop_idx < _num_loops; loop_idx++) {
1128     int block_idx = 0;
1129     assert(!is_block_in_loop(loop_idx, _linear_scan_order->at(block_idx)), "the first block must not be present in any loop");
1130 
1131     // skip blocks before the loop
1132     while (block_idx < _num_blocks && !is_block_in_loop(loop_idx, _linear_scan_order->at(block_idx))) {
1133       block_idx++;
1134     }
1135     // skip blocks of loop
1136     while (block_idx < _num_blocks && is_block_in_loop(loop_idx, _linear_scan_order->at(block_idx))) {
1137       block_idx++;
1138     }
1139     // after the first non-loop block, there must not be another loop-block
1140     while (block_idx < _num_blocks) {
1141       assert(!is_block_in_loop(loop_idx, _linear_scan_order->at(block_idx)), "loop not continuous in linear-scan order");
1142       block_idx++;
1143     }
1144   }
1145 }
1146 #endif
1147 
1148 
1149 void IR::compute_code() {
1150   assert(is_valid(), "IR must be valid");
1151 
1152   ComputeLinearScanOrder compute_order(compilation(), start());
1153   _num_loops = compute_order.num_loops();
1154   _code = compute_order.linear_scan_order();
1155 }
1156 
1157 
1158 void IR::compute_use_counts() {
1159   // make sure all values coming out of this block get evaluated.
1160   int num_blocks = _code->length();
1161   for (int i = 0; i < num_blocks; i++) {
1162     _code->at(i)->end()->state()->pin_stack_for_linear_scan();
1163   }
1164 
1165   // compute use counts
1166   UseCountComputer::compute(_code);
1167 }
1168 
1169 
1170 void IR::iterate_preorder(BlockClosure* closure) {
1171   assert(is_valid(), "IR must be valid");
1172   start()->iterate_preorder(closure);
1173 }
1174 
1175 
1176 void IR::iterate_postorder(BlockClosure* closure) {
1177   assert(is_valid(), "IR must be valid");
1178   start()->iterate_postorder(closure);
1179 }
1180 
1181 void IR::iterate_linear_scan_order(BlockClosure* closure) {
1182   linear_scan_order()->iterate_forward(closure);
1183 }
1184 
1185 
1186 #ifndef PRODUCT
1187 class BlockPrinter: public BlockClosure {
1188  private:
1189   InstructionPrinter* _ip;
1190   bool                _cfg_only;
1191   bool                _live_only;
1192 
1193  public:
1194   BlockPrinter(InstructionPrinter* ip, bool cfg_only, bool live_only = false) {
1195     _ip       = ip;
1196     _cfg_only = cfg_only;
1197     _live_only = live_only;
1198   }
1199 
1200   virtual void block_do(BlockBegin* block) {
1201     if (_cfg_only) {
1202       _ip->print_instr(block); tty->cr();
1203     } else {
1204       block->print_block(*_ip, _live_only);
1205     }
1206   }
1207 };
1208 
1209 
1210 void IR::print(BlockBegin* start, bool cfg_only, bool live_only) {
1211   ttyLocker ttyl;
1212   InstructionPrinter ip(!cfg_only);
1213   BlockPrinter bp(&ip, cfg_only, live_only);
1214   start->iterate_preorder(&bp);
1215   tty->cr();
1216 }
1217 
1218 void IR::print(bool cfg_only, bool live_only) {
1219   if (is_valid()) {
1220     print(start(), cfg_only, live_only);
1221   } else {
1222     tty->print_cr("invalid IR");
1223   }
1224 }
1225 
1226 
1227 define_array(BlockListArray, BlockList*)
1228 define_stack(BlockListList, BlockListArray)
1229 
1230 class PredecessorValidator : public BlockClosure {
1231  private:
1232   BlockListList* _predecessors;
1233   BlockList*     _blocks;
1234 
1235   static int cmp(BlockBegin** a, BlockBegin** b) {
1236     return (*a)->block_id() - (*b)->block_id();
1237   }
1238 
1239  public:
1240   PredecessorValidator(IR* hir) {
1241     ResourceMark rm;
1242     _predecessors = new BlockListList(BlockBegin::number_of_blocks(), NULL);
1243     _blocks = new BlockList();
1244 
1245     int i;
1246     hir->start()->iterate_preorder(this);
1247     if (hir->code() != NULL) {
1248       assert(hir->code()->length() == _blocks->length(), "must match");
1249       for (i = 0; i < _blocks->length(); i++) {
1250         assert(hir->code()->contains(_blocks->at(i)), "should be in both lists");
1251       }
1252     }
1253 
1254     for (i = 0; i < _blocks->length(); i++) {
1255       BlockBegin* block = _blocks->at(i);
1256       BlockList* preds = _predecessors->at(block->block_id());
1257       if (preds == NULL) {
1258         assert(block->number_of_preds() == 0, "should be the same");
1259         continue;
1260       }
1261 
1262       // clone the pred list so we can mutate it
1263       BlockList* pred_copy = new BlockList();
1264       int j;
1265       for (j = 0; j < block->number_of_preds(); j++) {
1266         pred_copy->append(block->pred_at(j));
1267       }
1268       // sort them in the same order
1269       preds->sort(cmp);
1270       pred_copy->sort(cmp);
1271       int length = MIN2(preds->length(), block->number_of_preds());
1272       for (j = 0; j < block->number_of_preds(); j++) {
1273         assert(preds->at(j) == pred_copy->at(j), "must match");
1274       }
1275 
1276       assert(preds->length() == block->number_of_preds(), "should be the same");
1277     }
1278   }
1279 
1280   virtual void block_do(BlockBegin* block) {
1281     _blocks->append(block);
1282     BlockEnd* be = block->end();
1283     int n = be->number_of_sux();
1284     int i;
1285     for (i = 0; i < n; i++) {
1286       BlockBegin* sux = be->sux_at(i);
1287       assert(!sux->is_set(BlockBegin::exception_entry_flag), "must not be xhandler");
1288 
1289       BlockList* preds = _predecessors->at_grow(sux->block_id(), NULL);
1290       if (preds == NULL) {
1291         preds = new BlockList();
1292         _predecessors->at_put(sux->block_id(), preds);
1293       }
1294       preds->append(block);
1295     }
1296 
1297     n = block->number_of_exception_handlers();
1298     for (i = 0; i < n; i++) {
1299       BlockBegin* sux = block->exception_handler_at(i);
1300       assert(sux->is_set(BlockBegin::exception_entry_flag), "must be xhandler");
1301 
1302       BlockList* preds = _predecessors->at_grow(sux->block_id(), NULL);
1303       if (preds == NULL) {
1304         preds = new BlockList();
1305         _predecessors->at_put(sux->block_id(), preds);
1306       }
1307       preds->append(block);
1308     }
1309   }
1310 };
1311 
1312 class VerifyBlockBeginField : public BlockClosure {
1313 
1314 public:
1315 
1316   virtual void block_do(BlockBegin *block) {
1317 
1318     Instruction *cur = block;
1319     while (cur) {
1320       assert(cur->block() == block, "Block begin is not correct");
1321       cur = cur->next();
1322     }
1323   }
1324 };
1325 
1326 void IR::verify() {
1327 #ifdef ASSERT
1328   PredecessorValidator pv(this);
1329   VerifyBlockBeginField verifier;
1330   this->iterate_postorder(&verifier);
1331 #endif
1332 }
1333 
1334 #endif // PRODUCT
1335 
1336 void SubstitutionResolver::visit(Value* v) {
1337   Value v0 = *v;
1338   if (v0) {
1339     Value vs = v0->subst();
1340     if (vs != v0) {
1341       *v = v0->subst();
1342     }
1343   }
1344 }
1345 
1346 #ifdef ASSERT
1347 class SubstitutionChecker: public ValueVisitor {
1348   void visit(Value* v) {
1349     Value v0 = *v;
1350     if (v0) {
1351       Value vs = v0->subst();
1352       assert(vs == v0, "missed substitution");
1353     }
1354   }
1355 };
1356 #endif
1357 
1358 
1359 void SubstitutionResolver::block_do(BlockBegin* block) {
1360   Instruction* last = NULL;
1361   for (Instruction* n = block; n != NULL;) {
1362     n->values_do(this);
1363     // need to remove this instruction from the instruction stream
1364     if (n->subst() != n) {
1365       assert(last != NULL, "must have last");
1366       last->set_next(n->next());
1367     } else {
1368       last = n;
1369     }
1370     n = last->next();
1371   }
1372 
1373 #ifdef ASSERT
1374   SubstitutionChecker check_substitute;
1375   if (block->state()) block->state()->values_do(&check_substitute);
1376   block->block_values_do(&check_substitute);
1377   if (block->end() && block->end()->state()) block->end()->state()->values_do(&check_substitute);
1378 #endif
1379 }