Dale Johannesen | 72f1596 | 2007-07-13 17:31:29 +0000 | [diff] [blame] | 1 | //===----- SchedulePostRAList.cpp - list scheduler ------------------------===// |
Dale Johannesen | e7e7d0d | 2007-07-13 17:13:54 +0000 | [diff] [blame] | 2 | // |
| 3 | // The LLVM Compiler Infrastructure |
| 4 | // |
Chris Lattner | 4ee451d | 2007-12-29 20:36:04 +0000 | [diff] [blame] | 5 | // This file is distributed under the University of Illinois Open Source |
| 6 | // License. See LICENSE.TXT for details. |
Dale Johannesen | e7e7d0d | 2007-07-13 17:13:54 +0000 | [diff] [blame] | 7 | // |
| 8 | //===----------------------------------------------------------------------===// |
| 9 | // |
| 10 | // This implements a top-down list scheduler, using standard algorithms. |
| 11 | // The basic approach uses a priority queue of available nodes to schedule. |
| 12 | // One at a time, nodes are taken from the priority queue (thus in priority |
| 13 | // order), checked for legality to schedule, and emitted if legal. |
| 14 | // |
| 15 | // Nodes may not be legal to schedule either due to structural hazards (e.g. |
| 16 | // pipeline or resource constraints) or because an input to the instruction has |
| 17 | // not completed execution. |
| 18 | // |
| 19 | //===----------------------------------------------------------------------===// |
| 20 | |
| 21 | #define DEBUG_TYPE "post-RA-sched" |
| 22 | #include "llvm/CodeGen/Passes.h" |
Dan Gohman | 343f0c0 | 2008-11-19 23:18:57 +0000 | [diff] [blame] | 23 | #include "llvm/CodeGen/ScheduleDAGInstrs.h" |
| 24 | #include "llvm/CodeGen/LatencyPriorityQueue.h" |
| 25 | #include "llvm/CodeGen/SchedulerRegistry.h" |
Dale Johannesen | e7e7d0d | 2007-07-13 17:13:54 +0000 | [diff] [blame] | 26 | #include "llvm/CodeGen/MachineFunctionPass.h" |
Dan Gohman | 21d9003 | 2008-11-25 00:52:40 +0000 | [diff] [blame^] | 27 | #include "llvm/CodeGen/MachineRegisterInfo.h" |
| 28 | #include "llvm/Target/TargetInstrInfo.h" |
| 29 | #include "llvm/Target/TargetRegisterInfo.h" |
Chris Lattner | 459525d | 2008-01-14 19:00:06 +0000 | [diff] [blame] | 30 | #include "llvm/Support/Compiler.h" |
Dale Johannesen | e7e7d0d | 2007-07-13 17:13:54 +0000 | [diff] [blame] | 31 | #include "llvm/Support/Debug.h" |
Dan Gohman | 343f0c0 | 2008-11-19 23:18:57 +0000 | [diff] [blame] | 32 | #include "llvm/ADT/Statistic.h" |
Dan Gohman | 21d9003 | 2008-11-25 00:52:40 +0000 | [diff] [blame^] | 33 | #include "llvm/ADT/DenseSet.h" |
| 34 | #include <map> |
| 35 | #include <climits> |
Dale Johannesen | e7e7d0d | 2007-07-13 17:13:54 +0000 | [diff] [blame] | 36 | using namespace llvm; |
| 37 | |
Dan Gohman | 343f0c0 | 2008-11-19 23:18:57 +0000 | [diff] [blame] | 38 | STATISTIC(NumStalls, "Number of pipeline stalls"); |
| 39 | |
Dan Gohman | 21d9003 | 2008-11-25 00:52:40 +0000 | [diff] [blame^] | 40 | static cl::opt<bool> |
| 41 | EnableAntiDepBreaking("break-anti-dependencies", |
| 42 | cl::desc("Break scheduling anti-dependencies"), |
| 43 | cl::init(false)); |
| 44 | |
Dale Johannesen | e7e7d0d | 2007-07-13 17:13:54 +0000 | [diff] [blame] | 45 | namespace { |
Dan Gohman | 343f0c0 | 2008-11-19 23:18:57 +0000 | [diff] [blame] | 46 | class VISIBILITY_HIDDEN PostRAScheduler : public MachineFunctionPass { |
Dale Johannesen | e7e7d0d | 2007-07-13 17:13:54 +0000 | [diff] [blame] | 47 | public: |
| 48 | static char ID; |
Dan Gohman | 343f0c0 | 2008-11-19 23:18:57 +0000 | [diff] [blame] | 49 | PostRAScheduler() : MachineFunctionPass(&ID) {} |
Dan Gohman | 21d9003 | 2008-11-25 00:52:40 +0000 | [diff] [blame^] | 50 | |
Dale Johannesen | e7e7d0d | 2007-07-13 17:13:54 +0000 | [diff] [blame] | 51 | const char *getPassName() const { |
Dan Gohman | 21d9003 | 2008-11-25 00:52:40 +0000 | [diff] [blame^] | 52 | return "Post RA top-down list latency scheduler"; |
Dale Johannesen | e7e7d0d | 2007-07-13 17:13:54 +0000 | [diff] [blame] | 53 | } |
| 54 | |
| 55 | bool runOnMachineFunction(MachineFunction &Fn); |
| 56 | }; |
Dan Gohman | 343f0c0 | 2008-11-19 23:18:57 +0000 | [diff] [blame] | 57 | char PostRAScheduler::ID = 0; |
| 58 | |
| 59 | class VISIBILITY_HIDDEN SchedulePostRATDList : public ScheduleDAGInstrs { |
Dan Gohman | 343f0c0 | 2008-11-19 23:18:57 +0000 | [diff] [blame] | 60 | /// AvailableQueue - The priority queue to use for the available SUnits. |
| 61 | /// |
| 62 | LatencyPriorityQueue AvailableQueue; |
| 63 | |
| 64 | /// PendingQueue - This contains all of the instructions whose operands have |
| 65 | /// been issued, but their results are not ready yet (due to the latency of |
| 66 | /// the operation). Once the operands becomes available, the instruction is |
| 67 | /// added to the AvailableQueue. |
| 68 | std::vector<SUnit*> PendingQueue; |
| 69 | |
Dan Gohman | 21d9003 | 2008-11-25 00:52:40 +0000 | [diff] [blame^] | 70 | /// Topo - A topological ordering for SUnits. |
| 71 | ScheduleDAGTopologicalSort Topo; |
Dan Gohman | 343f0c0 | 2008-11-19 23:18:57 +0000 | [diff] [blame] | 72 | |
Dan Gohman | 21d9003 | 2008-11-25 00:52:40 +0000 | [diff] [blame^] | 73 | public: |
| 74 | SchedulePostRATDList(MachineBasicBlock *mbb, const TargetMachine &tm) |
| 75 | : ScheduleDAGInstrs(mbb, tm), Topo(SUnits) {} |
Dan Gohman | 343f0c0 | 2008-11-19 23:18:57 +0000 | [diff] [blame] | 76 | |
| 77 | void Schedule(); |
| 78 | |
| 79 | private: |
| 80 | void ReleaseSucc(SUnit *SU, SUnit *SuccSU, bool isChain); |
| 81 | void ScheduleNodeTopDown(SUnit *SU, unsigned CurCycle); |
| 82 | void ListScheduleTopDown(); |
Dan Gohman | 21d9003 | 2008-11-25 00:52:40 +0000 | [diff] [blame^] | 83 | bool BreakAntiDependencies(); |
Dan Gohman | 343f0c0 | 2008-11-19 23:18:57 +0000 | [diff] [blame] | 84 | }; |
Dale Johannesen | e7e7d0d | 2007-07-13 17:13:54 +0000 | [diff] [blame] | 85 | } |
| 86 | |
Dan Gohman | 343f0c0 | 2008-11-19 23:18:57 +0000 | [diff] [blame] | 87 | bool PostRAScheduler::runOnMachineFunction(MachineFunction &Fn) { |
| 88 | DOUT << "PostRAScheduler\n"; |
Dale Johannesen | e7e7d0d | 2007-07-13 17:13:54 +0000 | [diff] [blame] | 89 | |
| 90 | // Loop over all of the basic blocks |
| 91 | for (MachineFunction::iterator MBB = Fn.begin(), MBBe = Fn.end(); |
Dan Gohman | 343f0c0 | 2008-11-19 23:18:57 +0000 | [diff] [blame] | 92 | MBB != MBBe; ++MBB) { |
| 93 | |
Dan Gohman | 21d9003 | 2008-11-25 00:52:40 +0000 | [diff] [blame^] | 94 | SchedulePostRATDList Scheduler(MBB, Fn.getTarget()); |
Dan Gohman | 343f0c0 | 2008-11-19 23:18:57 +0000 | [diff] [blame] | 95 | |
| 96 | Scheduler.Run(); |
| 97 | |
| 98 | Scheduler.EmitSchedule(); |
| 99 | } |
Dale Johannesen | e7e7d0d | 2007-07-13 17:13:54 +0000 | [diff] [blame] | 100 | |
| 101 | return true; |
| 102 | } |
| 103 | |
Dan Gohman | 343f0c0 | 2008-11-19 23:18:57 +0000 | [diff] [blame] | 104 | /// Schedule - Schedule the DAG using list scheduling. |
| 105 | void SchedulePostRATDList::Schedule() { |
| 106 | DOUT << "********** List Scheduling **********\n"; |
| 107 | |
| 108 | // Build scheduling units. |
| 109 | BuildSchedUnits(); |
| 110 | |
Dan Gohman | 21d9003 | 2008-11-25 00:52:40 +0000 | [diff] [blame^] | 111 | if (EnableAntiDepBreaking) { |
| 112 | if (BreakAntiDependencies()) { |
| 113 | // We made changes. Update the dependency graph. |
| 114 | // Theoretically we could update the graph in place: |
| 115 | // When a live range is changed to use a different register, remove |
| 116 | // the def's anti-dependence *and* output-dependence edges due to |
| 117 | // that register, and add new anti-dependence and output-dependence |
| 118 | // edges based on the next live range of the register. |
| 119 | SUnits.clear(); |
| 120 | BuildSchedUnits(); |
| 121 | } |
| 122 | } |
| 123 | |
Dan Gohman | 343f0c0 | 2008-11-19 23:18:57 +0000 | [diff] [blame] | 124 | AvailableQueue.initNodes(SUnits); |
Dan Gohman | 21d9003 | 2008-11-25 00:52:40 +0000 | [diff] [blame^] | 125 | |
Dan Gohman | 343f0c0 | 2008-11-19 23:18:57 +0000 | [diff] [blame] | 126 | ListScheduleTopDown(); |
| 127 | |
| 128 | AvailableQueue.releaseState(); |
| 129 | } |
| 130 | |
Dan Gohman | 21d9003 | 2008-11-25 00:52:40 +0000 | [diff] [blame^] | 131 | /// getInstrOperandRegClass - Return register class of the operand of an |
| 132 | /// instruction of the specified TargetInstrDesc. |
| 133 | static const TargetRegisterClass* |
| 134 | getInstrOperandRegClass(const TargetRegisterInfo *TRI, |
| 135 | const TargetInstrInfo *TII, const TargetInstrDesc &II, |
| 136 | unsigned Op) { |
| 137 | if (Op >= II.getNumOperands()) |
| 138 | return NULL; |
| 139 | if (II.OpInfo[Op].isLookupPtrRegClass()) |
| 140 | return TII->getPointerRegClass(); |
| 141 | return TRI->getRegClass(II.OpInfo[Op].RegClass); |
| 142 | } |
| 143 | |
| 144 | /// BreakAntiDependencies - Identifiy anti-dependencies along the critical path |
| 145 | /// of the ScheduleDAG and break them by renaming registers. |
| 146 | /// |
| 147 | bool SchedulePostRATDList::BreakAntiDependencies() { |
| 148 | // The code below assumes that there is at least one instruction, |
| 149 | // so just duck out immediately if the block is empty. |
| 150 | if (BB->empty()) return false; |
| 151 | |
| 152 | Topo.InitDAGTopologicalSorting(); |
| 153 | |
| 154 | // Compute a critical path for the DAG. |
| 155 | SUnit *Max = 0; |
| 156 | std::vector<SDep *> CriticalPath(SUnits.size()); |
| 157 | for (ScheduleDAGTopologicalSort::const_iterator I = Topo.begin(), |
| 158 | E = Topo.end(); I != E; ++I) { |
| 159 | SUnit *SU = &SUnits[*I]; |
| 160 | for (SUnit::pred_iterator P = SU->Preds.begin(), PE = SU->Preds.end(); |
| 161 | P != PE; ++P) { |
| 162 | SUnit *PredSU = P->Dep; |
| 163 | unsigned PredLatency = PredSU->CycleBound + PredSU->Latency; |
| 164 | if (SU->CycleBound < PredLatency) { |
| 165 | SU->CycleBound = PredLatency; |
| 166 | CriticalPath[*I] = &*P; |
| 167 | } |
| 168 | } |
| 169 | // Keep track of the node at the end of the critical path. |
| 170 | if (!Max || SU->CycleBound + SU->Latency > Max->CycleBound + Max->Latency) |
| 171 | Max = SU; |
| 172 | } |
| 173 | |
| 174 | DOUT << "Critical path has total latency " |
| 175 | << (Max ? Max->CycleBound + Max->Latency : 0) << "\n"; |
| 176 | |
| 177 | // Walk the critical path from the bottom up. Collect all anti-dependence |
| 178 | // edges on the critical path. Skip anti-dependencies between SUnits that |
| 179 | // are connected with other edges, since such units won't be able to be |
| 180 | // scheduled past each other anyway. |
| 181 | // |
| 182 | // The heuristic is that edges on the critical path are more important to |
| 183 | // break than other edges. And since there are a limited number of |
| 184 | // registers, we don't want to waste them breaking edges that aren't |
| 185 | // important. |
| 186 | // |
| 187 | // TODO: Instructions with multiple defs could have multiple |
| 188 | // anti-dependencies. The current code here only knows how to break one |
| 189 | // edge per instruction. Note that we'd have to be able to break all of |
| 190 | // the anti-dependencies in an instruction in order to be effective. |
| 191 | BitVector AllocatableSet = TRI->getAllocatableSet(*MF); |
| 192 | DenseMap<MachineInstr *, unsigned> CriticalAntiDeps; |
| 193 | for (SUnit *SU = Max; CriticalPath[SU->NodeNum]; |
| 194 | SU = CriticalPath[SU->NodeNum]->Dep) { |
| 195 | SDep *Edge = CriticalPath[SU->NodeNum]; |
| 196 | SUnit *NextSU = Edge->Dep; |
| 197 | unsigned AntiDepReg = Edge->Reg; |
| 198 | // Don't break anti-dependencies on non-allocatable registers. |
| 199 | if (!AllocatableSet.test(AntiDepReg)) |
| 200 | continue; |
| 201 | // If the SUnit has other dependencies on the SUnit that it |
| 202 | // anti-depends on, don't bother breaking the anti-dependency. |
| 203 | // Also, if there are dependencies on other SUnits with the |
| 204 | // same register as the anti-dependency, don't attempt to |
| 205 | // break it. |
| 206 | for (SUnit::pred_iterator P = SU->Preds.begin(), PE = SU->Preds.end(); |
| 207 | P != PE; ++P) |
| 208 | if (P->Dep == NextSU ? |
| 209 | (!P->isAntiDep || P->Reg != AntiDepReg) : |
| 210 | (!P->isCtrl && !P->isAntiDep && P->Reg == AntiDepReg)) { |
| 211 | AntiDepReg = 0; |
| 212 | break; |
| 213 | } |
| 214 | if (AntiDepReg != 0) |
| 215 | CriticalAntiDeps[SU->getInstr()] = AntiDepReg; |
| 216 | } |
| 217 | |
| 218 | // For live regs that are only used in one register class in a live range, |
| 219 | // the register class. If the register is not live or is referenced in |
| 220 | // multiple register classes, the corresponding value is null. If the |
| 221 | // register is used in multiple register classes, the corresponding value |
| 222 | // is -1 casted to a pointer. |
| 223 | const TargetRegisterClass * |
| 224 | Classes[TargetRegisterInfo::FirstVirtualRegister] = {}; |
| 225 | |
| 226 | // Map registers to all their references within a live range. |
| 227 | std::multimap<unsigned, MachineOperand *> RegRefs; |
| 228 | |
| 229 | // The index of the most recent kill (proceding bottom-up), or -1 if |
| 230 | // the register is not live. |
| 231 | unsigned KillIndices[TargetRegisterInfo::FirstVirtualRegister]; |
| 232 | std::fill(KillIndices, array_endof(KillIndices), -1); |
| 233 | // The index of the most recent def (proceding bottom up), or -1 if |
| 234 | // the register is live. |
| 235 | unsigned DefIndices[TargetRegisterInfo::FirstVirtualRegister]; |
| 236 | std::fill(DefIndices, array_endof(DefIndices), BB->size()); |
| 237 | |
| 238 | // Determine the live-out physregs for this block. |
| 239 | if (!BB->empty() && BB->back().getDesc().isReturn()) |
| 240 | // In a return block, examine the function live-out regs. |
| 241 | for (MachineRegisterInfo::liveout_iterator I = MRI.liveout_begin(), |
| 242 | E = MRI.liveout_end(); I != E; ++I) { |
| 243 | unsigned Reg = *I; |
| 244 | Classes[Reg] = reinterpret_cast<TargetRegisterClass *>(-1); |
| 245 | KillIndices[Reg] = BB->size(); |
| 246 | DefIndices[Reg] = -1; |
| 247 | // Repeat, for all aliases. |
| 248 | for (const unsigned *Alias = TRI->getAliasSet(Reg); *Alias; ++Alias) { |
| 249 | unsigned AliasReg = *Alias; |
| 250 | Classes[AliasReg] = reinterpret_cast<TargetRegisterClass *>(-1); |
| 251 | KillIndices[AliasReg] = BB->size(); |
| 252 | DefIndices[AliasReg] = -1; |
| 253 | } |
| 254 | } |
| 255 | else |
| 256 | // In a non-return block, examine the live-in regs of all successors. |
| 257 | for (MachineBasicBlock::succ_iterator SI = BB->succ_begin(), |
| 258 | SE = BB->succ_end(); SI != SE; ++SI) |
| 259 | for (MachineBasicBlock::livein_iterator I = (*SI)->livein_begin(), |
| 260 | E = (*SI)->livein_end(); I != E; ++I) { |
| 261 | unsigned Reg = *I; |
| 262 | Classes[Reg] = reinterpret_cast<TargetRegisterClass *>(-1); |
| 263 | KillIndices[Reg] = BB->size(); |
| 264 | DefIndices[Reg] = -1; |
| 265 | // Repeat, for all aliases. |
| 266 | for (const unsigned *Alias = TRI->getAliasSet(Reg); *Alias; ++Alias) { |
| 267 | unsigned AliasReg = *Alias; |
| 268 | Classes[AliasReg] = reinterpret_cast<TargetRegisterClass *>(-1); |
| 269 | KillIndices[AliasReg] = BB->size(); |
| 270 | DefIndices[AliasReg] = -1; |
| 271 | } |
| 272 | } |
| 273 | |
| 274 | // Consider callee-saved registers as live-out, since we're running after |
| 275 | // prologue/epilogue insertion so there's no way to add additional |
| 276 | // saved registers. |
| 277 | // |
| 278 | // TODO: If the callee saves and restores these, then we can potentially |
| 279 | // use them between the save and the restore. To do that, we could scan |
| 280 | // the exit blocks to see which of these registers are defined. |
| 281 | for (const unsigned *I = TRI->getCalleeSavedRegs(); *I; ++I) { |
| 282 | unsigned Reg = *I; |
| 283 | Classes[Reg] = reinterpret_cast<TargetRegisterClass *>(-1); |
| 284 | KillIndices[Reg] = BB->size(); |
| 285 | DefIndices[Reg] = -1; |
| 286 | // Repeat, for all aliases. |
| 287 | for (const unsigned *Alias = TRI->getAliasSet(Reg); *Alias; ++Alias) { |
| 288 | unsigned AliasReg = *Alias; |
| 289 | Classes[AliasReg] = reinterpret_cast<TargetRegisterClass *>(-1); |
| 290 | KillIndices[AliasReg] = BB->size(); |
| 291 | DefIndices[AliasReg] = -1; |
| 292 | } |
| 293 | } |
| 294 | |
| 295 | // Consider this pattern: |
| 296 | // A = ... |
| 297 | // ... = A |
| 298 | // A = ... |
| 299 | // ... = A |
| 300 | // A = ... |
| 301 | // ... = A |
| 302 | // A = ... |
| 303 | // ... = A |
| 304 | // There are three anti-dependencies here, and without special care, |
| 305 | // we'd break all of them using the same register: |
| 306 | // A = ... |
| 307 | // ... = A |
| 308 | // B = ... |
| 309 | // ... = B |
| 310 | // B = ... |
| 311 | // ... = B |
| 312 | // B = ... |
| 313 | // ... = B |
| 314 | // because at each anti-dependence, B is the first register that |
| 315 | // isn't A which is free. This re-introduces anti-dependencies |
| 316 | // at all but one of the original anti-dependencies that we were |
| 317 | // trying to break. To avoid this, keep track of the most recent |
| 318 | // register that each register was replaced with, avoid avoid |
| 319 | // using it to repair an anti-dependence on the same register. |
| 320 | // This lets us produce this: |
| 321 | // A = ... |
| 322 | // ... = A |
| 323 | // B = ... |
| 324 | // ... = B |
| 325 | // C = ... |
| 326 | // ... = C |
| 327 | // B = ... |
| 328 | // ... = B |
| 329 | // This still has an anti-dependence on B, but at least it isn't on the |
| 330 | // original critical path. |
| 331 | // |
| 332 | // TODO: If we tracked more than one register here, we could potentially |
| 333 | // fix that remaining critical edge too. This is a little more involved, |
| 334 | // because unlike the most recent register, less recent registers should |
| 335 | // still be considered, though only if no other registers are available. |
| 336 | unsigned LastNewReg[TargetRegisterInfo::FirstVirtualRegister] = {}; |
| 337 | |
| 338 | // A registers defined and not used in an instruction. This is used for |
| 339 | // liveness tracking and is declared outside the loop only to avoid |
| 340 | // having it be re-allocated on each iteration. |
| 341 | DenseSet<unsigned> Defs; |
| 342 | |
| 343 | // Attempt to break anti-dependence edges on the critical path. Walk the |
| 344 | // instructions from the bottom up, tracking information about liveness |
| 345 | // as we go to help determine which registers are available. |
| 346 | bool Changed = false; |
| 347 | unsigned Count = BB->size() - 1; |
| 348 | for (MachineBasicBlock::reverse_iterator I = BB->rbegin(), E = BB->rend(); |
| 349 | I != E; ++I, --Count) { |
| 350 | MachineInstr *MI = &*I; |
| 351 | |
| 352 | // Check if this instruction has an anti-dependence that we're |
| 353 | // interested in. |
| 354 | DenseMap<MachineInstr *, unsigned>::iterator C = CriticalAntiDeps.find(MI); |
| 355 | unsigned AntiDepReg = C != CriticalAntiDeps.end() ? |
| 356 | C->second : 0; |
| 357 | |
| 358 | // Scan the register operands for this instruction and update |
| 359 | // Classes and RegRefs. |
| 360 | for (unsigned i = 0, e = MI->getNumOperands(); i != e; ++i) { |
| 361 | MachineOperand &MO = MI->getOperand(i); |
| 362 | if (!MO.isReg()) continue; |
| 363 | unsigned Reg = MO.getReg(); |
| 364 | if (Reg == 0) continue; |
| 365 | const TargetRegisterClass *NewRC = |
| 366 | getInstrOperandRegClass(TRI, TII, MI->getDesc(), i); |
| 367 | |
| 368 | // If this instruction has a use of AntiDepReg, breaking it |
| 369 | // is invalid. |
| 370 | if (MO.isUse() && AntiDepReg == Reg) |
| 371 | AntiDepReg = 0; |
| 372 | |
| 373 | // For now, only allow the register to be changed if its register |
| 374 | // class is consistent across all uses. |
| 375 | if (!Classes[Reg] && NewRC) |
| 376 | Classes[Reg] = NewRC; |
| 377 | else if (!NewRC || Classes[Reg] != NewRC) |
| 378 | Classes[Reg] = reinterpret_cast<TargetRegisterClass *>(-1); |
| 379 | |
| 380 | // Now check for aliases. |
| 381 | for (const unsigned *Alias = TRI->getAliasSet(Reg); *Alias; ++Alias) { |
| 382 | // If an alias of the reg is used during the live range, give up. |
| 383 | // Note that this allows us to skip checking if AntiDepReg |
| 384 | // overlaps with any of the aliases, among other things. |
| 385 | unsigned AliasReg = *Alias; |
| 386 | if (Classes[AliasReg]) { |
| 387 | Classes[AliasReg] = reinterpret_cast<TargetRegisterClass *>(-1); |
| 388 | Classes[Reg] = reinterpret_cast<TargetRegisterClass *>(-1); |
| 389 | } |
| 390 | } |
| 391 | |
| 392 | // If we're still willing to consider this register, note the reference. |
| 393 | if (Classes[Reg] != reinterpret_cast<TargetRegisterClass *>(-1)) |
| 394 | RegRefs.insert(std::make_pair(Reg, &MO)); |
| 395 | } |
| 396 | |
| 397 | // Determine AntiDepReg's register class, if it is live and is |
| 398 | // consistently used within a single class. |
| 399 | const TargetRegisterClass *RC = AntiDepReg != 0 ? Classes[AntiDepReg] : 0; |
| 400 | assert(AntiDepReg == 0 || RC != NULL && |
| 401 | "Register should be live if it's causing an anti-dependence!"); |
| 402 | if (RC == reinterpret_cast<TargetRegisterClass *>(-1)) |
| 403 | AntiDepReg = 0; |
| 404 | |
| 405 | // Look for a suitable register to use to break the anti-depenence. |
| 406 | // |
| 407 | // TODO: Instead of picking the first free register, consider which might |
| 408 | // be the best. |
| 409 | if (AntiDepReg != 0) { |
| 410 | for (TargetRegisterClass::iterator R = RC->allocation_order_begin(*MF), |
| 411 | RE = RC->allocation_order_end(*MF); R != RE; ++R) { |
| 412 | unsigned NewReg = *R; |
| 413 | // Don't replace a register with itself. |
| 414 | if (NewReg == AntiDepReg) continue; |
| 415 | // Don't replace a register with one that was recently used to repair |
| 416 | // an anti-dependence with this AntiDepReg, because that would |
| 417 | // re-introduce that anti-dependence. |
| 418 | if (NewReg == LastNewReg[AntiDepReg]) continue; |
| 419 | // If NewReg is dead and NewReg's most recent def is not before |
| 420 | // AntiDepReg's kill, it's safe to replace AntiDepReg with NewReg. |
| 421 | assert(((KillIndices[AntiDepReg] == -1) != (DefIndices[AntiDepReg] == -1)) && |
| 422 | "Kill and Def maps aren't consistent for AntiDepReg!"); |
| 423 | assert(((KillIndices[NewReg] == -1) != (DefIndices[NewReg] == -1)) && |
| 424 | "Kill and Def maps aren't consistent for NewReg!"); |
| 425 | if (KillIndices[NewReg] == -1 && |
| 426 | KillIndices[AntiDepReg] <= DefIndices[NewReg]) { |
| 427 | DOUT << "Breaking anti-dependence edge on reg " << AntiDepReg |
| 428 | << " with reg " << NewReg << "!\n"; |
| 429 | |
| 430 | // Update the references to the old register to refer to the new |
| 431 | // register. |
| 432 | std::pair<std::multimap<unsigned, MachineOperand *>::iterator, |
| 433 | std::multimap<unsigned, MachineOperand *>::iterator> |
| 434 | Range = RegRefs.equal_range(AntiDepReg); |
| 435 | for (std::multimap<unsigned, MachineOperand *>::iterator |
| 436 | Q = Range.first, QE = Range.second; Q != QE; ++Q) |
| 437 | Q->second->setReg(NewReg); |
| 438 | |
| 439 | // We just went back in time and modified history; the |
| 440 | // liveness information for the anti-depenence reg is now |
| 441 | // inconsistent. Set the state as if it were dead. |
| 442 | Classes[NewReg] = Classes[AntiDepReg]; |
| 443 | DefIndices[NewReg] = DefIndices[AntiDepReg]; |
| 444 | KillIndices[NewReg] = KillIndices[AntiDepReg]; |
| 445 | |
| 446 | Classes[AntiDepReg] = 0; |
| 447 | DefIndices[AntiDepReg] = KillIndices[AntiDepReg]; |
| 448 | KillIndices[AntiDepReg] = -1; |
| 449 | |
| 450 | RegRefs.erase(AntiDepReg); |
| 451 | Changed = true; |
| 452 | LastNewReg[AntiDepReg] = NewReg; |
| 453 | break; |
| 454 | } |
| 455 | } |
| 456 | } |
| 457 | |
| 458 | // Update liveness. |
| 459 | Defs.clear(); |
| 460 | for (unsigned i = 0, e = MI->getNumOperands(); i != e; ++i) { |
| 461 | MachineOperand &MO = MI->getOperand(i); |
| 462 | if (!MO.isReg()) continue; |
| 463 | unsigned Reg = MO.getReg(); |
| 464 | if (Reg == 0) continue; |
| 465 | if (MO.isDef()) |
| 466 | Defs.insert(Reg); |
| 467 | else { |
| 468 | // Treat a use in the same instruction as a def as an extension of |
| 469 | // a live range. |
| 470 | Defs.erase(Reg); |
| 471 | // It wasn't previously live but now it is, this is a kill. |
| 472 | if (KillIndices[Reg] == -1) { |
| 473 | KillIndices[Reg] = Count; |
| 474 | DefIndices[Reg] = -1; |
| 475 | } |
| 476 | // Repeat, for all aliases. |
| 477 | for (const unsigned *Alias = TRI->getAliasSet(Reg); *Alias; ++Alias) { |
| 478 | unsigned AliasReg = *Alias; |
| 479 | Defs.erase(AliasReg); |
| 480 | if (KillIndices[AliasReg] == -1) { |
| 481 | KillIndices[AliasReg] = Count; |
| 482 | DefIndices[AliasReg] = -1; |
| 483 | } |
| 484 | } |
| 485 | } |
| 486 | } |
| 487 | // Proceding upwards, registers that are defed but not used in this |
| 488 | // instruction are now dead. |
| 489 | for (DenseSet<unsigned>::iterator D = Defs.begin(), DE = Defs.end(); |
| 490 | D != DE; ++D) { |
| 491 | unsigned Reg = *D; |
| 492 | DefIndices[Reg] = Count; |
| 493 | KillIndices[Reg] = -1; |
| 494 | Classes[Reg] = 0; |
| 495 | RegRefs.erase(Reg); |
| 496 | // Repeat, for all subregs. |
| 497 | for (const unsigned *Subreg = TRI->getSubRegisters(Reg); |
| 498 | *Subreg; ++Subreg) { |
| 499 | unsigned SubregReg = *Subreg; |
| 500 | DefIndices[SubregReg] = Count; |
| 501 | KillIndices[SubregReg] = -1; |
| 502 | Classes[SubregReg] = 0; |
| 503 | RegRefs.erase(SubregReg); |
| 504 | } |
| 505 | } |
| 506 | } |
| 507 | assert(Count == -1u && "Count mismatch!"); |
| 508 | |
| 509 | return Changed; |
| 510 | } |
| 511 | |
Dan Gohman | 343f0c0 | 2008-11-19 23:18:57 +0000 | [diff] [blame] | 512 | //===----------------------------------------------------------------------===// |
| 513 | // Top-Down Scheduling |
| 514 | //===----------------------------------------------------------------------===// |
| 515 | |
| 516 | /// ReleaseSucc - Decrement the NumPredsLeft count of a successor. Add it to |
| 517 | /// the PendingQueue if the count reaches zero. Also update its cycle bound. |
| 518 | void SchedulePostRATDList::ReleaseSucc(SUnit *SU, SUnit *SuccSU, bool isChain) { |
| 519 | --SuccSU->NumPredsLeft; |
| 520 | |
| 521 | #ifndef NDEBUG |
| 522 | if (SuccSU->NumPredsLeft < 0) { |
| 523 | cerr << "*** Scheduling failed! ***\n"; |
| 524 | SuccSU->dump(this); |
| 525 | cerr << " has been released too many times!\n"; |
| 526 | assert(0); |
| 527 | } |
| 528 | #endif |
| 529 | |
| 530 | // Compute how many cycles it will be before this actually becomes |
| 531 | // available. This is the max of the start time of all predecessors plus |
| 532 | // their latencies. |
| 533 | // If this is a token edge, we don't need to wait for the latency of the |
| 534 | // preceeding instruction (e.g. a long-latency load) unless there is also |
| 535 | // some other data dependence. |
| 536 | unsigned PredDoneCycle = SU->Cycle; |
| 537 | if (!isChain) |
| 538 | PredDoneCycle += SU->Latency; |
| 539 | else if (SU->Latency) |
| 540 | PredDoneCycle += 1; |
| 541 | SuccSU->CycleBound = std::max(SuccSU->CycleBound, PredDoneCycle); |
| 542 | |
| 543 | if (SuccSU->NumPredsLeft == 0) { |
| 544 | PendingQueue.push_back(SuccSU); |
| 545 | } |
| 546 | } |
| 547 | |
| 548 | /// ScheduleNodeTopDown - Add the node to the schedule. Decrement the pending |
| 549 | /// count of its successors. If a successor pending count is zero, add it to |
| 550 | /// the Available queue. |
| 551 | void SchedulePostRATDList::ScheduleNodeTopDown(SUnit *SU, unsigned CurCycle) { |
| 552 | DOUT << "*** Scheduling [" << CurCycle << "]: "; |
| 553 | DEBUG(SU->dump(this)); |
| 554 | |
| 555 | Sequence.push_back(SU); |
| 556 | SU->Cycle = CurCycle; |
| 557 | |
| 558 | // Top down: release successors. |
| 559 | for (SUnit::succ_iterator I = SU->Succs.begin(), E = SU->Succs.end(); |
| 560 | I != E; ++I) |
| 561 | ReleaseSucc(SU, I->Dep, I->isCtrl); |
| 562 | |
| 563 | SU->isScheduled = true; |
| 564 | AvailableQueue.ScheduledNode(SU); |
| 565 | } |
| 566 | |
| 567 | /// ListScheduleTopDown - The main loop of list scheduling for top-down |
| 568 | /// schedulers. |
| 569 | void SchedulePostRATDList::ListScheduleTopDown() { |
| 570 | unsigned CurCycle = 0; |
| 571 | |
| 572 | // All leaves to Available queue. |
| 573 | for (unsigned i = 0, e = SUnits.size(); i != e; ++i) { |
| 574 | // It is available if it has no predecessors. |
| 575 | if (SUnits[i].Preds.empty()) { |
| 576 | AvailableQueue.push(&SUnits[i]); |
| 577 | SUnits[i].isAvailable = true; |
| 578 | } |
| 579 | } |
| 580 | |
| 581 | // While Available queue is not empty, grab the node with the highest |
| 582 | // priority. If it is not ready put it back. Schedule the node. |
| 583 | Sequence.reserve(SUnits.size()); |
| 584 | while (!AvailableQueue.empty() || !PendingQueue.empty()) { |
| 585 | // Check to see if any of the pending instructions are ready to issue. If |
| 586 | // so, add them to the available queue. |
| 587 | for (unsigned i = 0, e = PendingQueue.size(); i != e; ++i) { |
| 588 | if (PendingQueue[i]->CycleBound == CurCycle) { |
| 589 | AvailableQueue.push(PendingQueue[i]); |
| 590 | PendingQueue[i]->isAvailable = true; |
| 591 | PendingQueue[i] = PendingQueue.back(); |
| 592 | PendingQueue.pop_back(); |
| 593 | --i; --e; |
| 594 | } else { |
| 595 | assert(PendingQueue[i]->CycleBound > CurCycle && "Negative latency?"); |
| 596 | } |
| 597 | } |
| 598 | |
Dan Gohman | 21d9003 | 2008-11-25 00:52:40 +0000 | [diff] [blame^] | 599 | // If there are no instructions available, don't try to issue anything. |
Dan Gohman | 343f0c0 | 2008-11-19 23:18:57 +0000 | [diff] [blame] | 600 | if (AvailableQueue.empty()) { |
| 601 | ++CurCycle; |
| 602 | continue; |
| 603 | } |
| 604 | |
| 605 | SUnit *FoundSUnit = AvailableQueue.pop(); |
| 606 | |
| 607 | // If we found a node to schedule, do it now. |
| 608 | if (FoundSUnit) { |
| 609 | ScheduleNodeTopDown(FoundSUnit, CurCycle); |
| 610 | |
| 611 | // If this is a pseudo-op node, we don't want to increment the current |
| 612 | // cycle. |
| 613 | if (FoundSUnit->Latency) // Don't increment CurCycle for pseudo-ops! |
| 614 | ++CurCycle; |
| 615 | } else { |
| 616 | // Otherwise, we have a pipeline stall, but no other problem, just advance |
| 617 | // the current cycle and try again. |
| 618 | DOUT << "*** Advancing cycle, no work to do\n"; |
| 619 | ++NumStalls; |
| 620 | ++CurCycle; |
| 621 | } |
| 622 | } |
| 623 | |
| 624 | #ifndef NDEBUG |
Dan Gohman | a1e6d36 | 2008-11-20 01:26:25 +0000 | [diff] [blame] | 625 | VerifySchedule(/*isBottomUp=*/false); |
Dan Gohman | 343f0c0 | 2008-11-19 23:18:57 +0000 | [diff] [blame] | 626 | #endif |
| 627 | } |
Dale Johannesen | e7e7d0d | 2007-07-13 17:13:54 +0000 | [diff] [blame] | 628 | |
| 629 | //===----------------------------------------------------------------------===// |
| 630 | // Public Constructor Functions |
| 631 | //===----------------------------------------------------------------------===// |
| 632 | |
| 633 | FunctionPass *llvm::createPostRAScheduler() { |
Dan Gohman | 343f0c0 | 2008-11-19 23:18:57 +0000 | [diff] [blame] | 634 | return new PostRAScheduler(); |
Dale Johannesen | e7e7d0d | 2007-07-13 17:13:54 +0000 | [diff] [blame] | 635 | } |