llvm-capstone/llvm/lib/CodeGen
Craig Topper 846429de74 [DAGCombiner][X86] Teach SimplifyVBinOp to fold VBinOp (concat X, undef/constant), (concat Y, undef/constant) -> concat (VBinOp X, Y), VecC
This improves the combine I included in D66504 to handle constants in the upper operands of the concat. If we can constant fold them away we can pull the concat after the bin op. This helps with chains of madd reductions on X86 from loop unrolling. The loop madd reduction pattern creates pmaddwd with half the width of the add that follows it using zeroes to fill the upper bits. If we have two of these added together we can pull the zeroes through the accumulating add and then shrink it.

Differential Revision: https://reviews.llvm.org/D66680

llvm-svn: 369937
2019-08-26 17:59:11 +00:00
..
AsmPrinter [PowerPC][AIX] Adds support for writing the .data section in assembly files 2019-08-25 15:17:25 +00:00
GlobalISel [GlobalISel] Legalizer: Retry combining illegal artifacts as long as there new artifacts 2019-08-23 20:30:35 +00:00
MIRParser [DebugInfo] Allow bundled calls in the MIR's call site info 2019-08-19 12:41:22 +00:00
SelectionDAG [DAGCombiner][X86] Teach SimplifyVBinOp to fold VBinOp (concat X, undef/constant), (concat Y, undef/constant) -> concat (VBinOp X, Y), VecC 2019-08-26 17:59:11 +00:00
AggressiveAntiDepBreaker.cpp Apply llvm-prefer-register-over-unsigned from clang-tidy to LLVM 2019-08-15 19:22:08 +00:00
AggressiveAntiDepBreaker.h Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
AllocationOrder.cpp Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
AllocationOrder.h Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
Analysis.cpp IR. Change strip* family of functions to not look through aliases. 2019-08-22 19:56:14 +00:00
AntiDepBreaker.h Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
AtomicExpandPass.cpp AtomicExpand: Don't crash on non-0 alloca 2019-06-11 01:35:07 +00:00
BasicTargetTransformInfo.cpp Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
BranchFolding.cpp Apply llvm-prefer-register-over-unsigned from clang-tidy to LLVM 2019-08-15 19:22:08 +00:00
BranchFolding.h Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
BranchRelaxation.cpp Escape % in printf format string. 2019-08-16 18:23:54 +00:00
BreakFalseDeps.cpp Apply llvm-prefer-register-over-unsigned from clang-tidy to LLVM 2019-08-15 19:22:08 +00:00
BuiltinGCs.cpp Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
CalcSpillWeights.cpp Apply llvm-prefer-register-over-unsigned from clang-tidy to LLVM 2019-08-15 19:22:08 +00:00
CallingConvLower.cpp [LLVM][Alignment] Introduce Alignment In CallingConv 2019-08-05 09:49:09 +00:00
CFIInstrInserter.cpp Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
CMakeLists.txt Revert "r364412 [ExpandMemCmp][MergeICmps] Move passes out of CodeGen into opt pipeline." 2019-06-26 12:13:13 +00:00
CodeGen.cpp [CodeGen] Add a pass to do block predication on SSA machine IR. 2019-08-20 15:54:59 +00:00
CodeGenPrepare.cpp [CGP] Remove ModifiedDT from the makeBitReverse loop 2019-08-19 18:02:24 +00:00
CriticalAntiDepBreaker.cpp Apply llvm-prefer-register-over-unsigned from clang-tidy to LLVM 2019-08-15 19:22:08 +00:00
CriticalAntiDepBreaker.h Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
DeadMachineInstructionElim.cpp Apply llvm-prefer-register-over-unsigned from clang-tidy to LLVM 2019-08-15 19:22:08 +00:00
DetectDeadLanes.cpp Apply llvm-prefer-register-over-unsigned from clang-tidy to LLVM 2019-08-15 19:22:08 +00:00
DFAPacketizer.cpp Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
DwarfEHPrepare.cpp Add an optional list of blocks to avoid when looking for a path in isPotentiallyReachable. 2019-04-02 01:05:48 +00:00
EarlyIfConversion.cpp [CodeGen] Add a pass to do block predication on SSA machine IR. 2019-08-20 15:54:59 +00:00
EdgeBundles.cpp Fix parameter name comments using clang-tidy. NFC. 2019-07-16 04:46:31 +00:00
ExecutionDomainFix.cpp Cleanup: llvm::bsearch -> llvm::partition_point after r364719 2019-06-30 11:19:56 +00:00
ExpandMemCmp.cpp Revert "r364412 [ExpandMemCmp][MergeICmps] Move passes out of CodeGen into opt pipeline." 2019-06-26 12:13:13 +00:00
ExpandPostRAPseudos.cpp Apply llvm-prefer-register-over-unsigned from clang-tidy to LLVM 2019-08-15 19:22:08 +00:00
ExpandReductions.cpp Change semantics of fadd/fmul vector reductions. 2019-06-11 08:22:10 +00:00
FaultMaps.cpp Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
FEntryInserter.cpp Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
FinalizeISel.cpp Rename ExpandISelPseudo->FinalizeISel, delay register reservation 2019-06-19 00:25:39 +00:00
FuncletLayout.cpp Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
GCMetadata.cpp [llvm] Migrate llvm::make_unique to std::make_unique 2019-08-15 15:54:37 +00:00
GCMetadataPrinter.cpp Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
GCRootLowering.cpp [opaque pointer types] Pass value type to LoadInst creation. 2019-02-01 20:44:24 +00:00
GCStrategy.cpp Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
GlobalMerge.cpp Use llvm::stable_sort 2019-04-23 14:51:27 +00:00
HardwareLoops.cpp Revert "[HardwareLoops] NFC - move hardware loop checking code to isHardwareLoopProfitable()" 2019-07-09 17:53:09 +00:00
IfConversion.cpp Apply llvm-prefer-register-over-unsigned from clang-tidy to LLVM 2019-08-15 19:22:08 +00:00
ImplicitNullChecks.cpp Apply llvm-prefer-register-over-unsigned from clang-tidy to LLVM 2019-08-15 19:22:08 +00:00
IndirectBrExpandPass.cpp Implementation of asm-goto support in LLVM 2019-02-08 20:48:56 +00:00
InlineSpiller.cpp Apply llvm-prefer-register-over-unsigned from clang-tidy to LLVM 2019-08-15 19:22:08 +00:00
InterferenceCache.cpp Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
InterferenceCache.h Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
InterleavedAccessPass.cpp [InterleavedAccessPass] Don't increase the number of bytes loaded. 2019-03-28 20:44:50 +00:00
InterleavedLoadCombinePass.cpp computePolynomialFromPointer - add missing early-out return for non-pointer types. 2019-04-29 19:25:16 +00:00
IntrinsicLowering.cpp [opaque pointer types] Pass function types to CallInst creation. 2019-02-01 20:43:25 +00:00
LatencyPriorityQueue.cpp Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
LazyMachineBlockFrequencyInfo.cpp [llvm] Migrate llvm::make_unique to std::make_unique 2019-08-15 15:54:37 +00:00
LexicalScopes.cpp Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
LiveDebugValues.cpp [DebugInfo] Remove invalidated locations during LiveDebugValues 2019-08-23 16:33:42 +00:00
LiveDebugVariables.cpp Apply llvm-prefer-register-over-unsigned from clang-tidy to LLVM 2019-08-15 19:22:08 +00:00
LiveDebugVariables.h Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
LiveInterval.cpp Finish moving TargetRegisterInfo::isVirtualRegister() and friends to llvm::Register as started by r367614. NFC 2019-08-01 23:27:28 +00:00
LiveIntervals.cpp Apply llvm-prefer-register-over-unsigned from clang-tidy to LLVM 2019-08-15 19:22:08 +00:00
LiveIntervalUnion.cpp Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
LivePhysRegs.cpp Apply llvm-prefer-register-over-unsigned from clang-tidy to LLVM 2019-08-15 19:22:08 +00:00
LiveRangeCalc.cpp Finish moving TargetRegisterInfo::isVirtualRegister() and friends to llvm::Register as started by r367614. NFC 2019-08-01 23:27:28 +00:00
LiveRangeCalc.h Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
LiveRangeEdit.cpp Apply llvm-prefer-register-over-unsigned from clang-tidy to LLVM 2019-08-15 19:22:08 +00:00
LiveRangeShrink.cpp Apply llvm-prefer-register-over-unsigned from clang-tidy to LLVM 2019-08-15 19:22:08 +00:00
LiveRangeUtils.h Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
LiveRegMatrix.cpp Apply llvm-prefer-register-over-unsigned from clang-tidy to LLVM 2019-08-15 19:22:08 +00:00
LiveRegUnits.cpp Apply llvm-prefer-register-over-unsigned from clang-tidy to LLVM 2019-08-15 19:22:08 +00:00
LiveStacks.cpp Finish moving TargetRegisterInfo::isVirtualRegister() and friends to llvm::Register as started by r367614. NFC 2019-08-01 23:27:28 +00:00
LiveVariables.cpp Apply llvm-prefer-register-over-unsigned from clang-tidy to LLVM 2019-08-15 19:22:08 +00:00
LLVMBuild.txt Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
LLVMTargetMachine.cpp [llvm] Migrate llvm::make_unique to std::make_unique 2019-08-15 15:54:37 +00:00
LocalStackSlotAllocation.cpp [LLVM][Alignment] Introduce Alignment In MachineFrameInfo 2019-08-21 14:29:30 +00:00
LoopTraversal.cpp Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
LowerEmuTLS.cpp Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
LowLevelType.cpp Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
MachineBasicBlock.cpp [SlotIndexes] Add print-slotindexes to disable printing slotindexes 2019-08-22 13:44:47 +00:00
MachineBlockFrequencyInfo.cpp Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
MachineBlockPlacement.cpp [MBP] Disable aggressive loop rotate in plain mode 2019-08-22 16:21:32 +00:00
MachineBranchProbabilityInfo.cpp Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
MachineCombiner.cpp Finish moving TargetRegisterInfo::isVirtualRegister() and friends to llvm::Register as started by r367614. NFC 2019-08-01 23:27:28 +00:00
MachineCopyPropagation.cpp Apply llvm-prefer-register-over-unsigned from clang-tidy to LLVM 2019-08-15 19:22:08 +00:00
MachineCSE.cpp Apply llvm-prefer-register-over-unsigned from clang-tidy to LLVM 2019-08-15 19:22:08 +00:00
MachineDominanceFrontier.cpp Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
MachineDominators.cpp Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
MachineFrameInfo.cpp [LLVM][Alignment] Introduce Alignment In MachineFrameInfo 2019-08-21 14:29:30 +00:00
MachineFunction.cpp [llvm] Migrate llvm::make_unique to std::make_unique 2019-08-15 15:54:37 +00:00
MachineFunctionPass.cpp Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
MachineFunctionPrinterPass.cpp Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
MachineInstr.cpp [llvm] Migrate llvm::make_unique to std::make_unique 2019-08-15 15:54:37 +00:00
MachineInstrBundle.cpp Apply llvm-prefer-register-over-unsigned from clang-tidy to LLVM 2019-08-15 19:22:08 +00:00
MachineLICM.cpp Apply llvm-prefer-register-over-unsigned from clang-tidy to LLVM 2019-08-15 19:22:08 +00:00
MachineLoopInfo.cpp Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
MachineModuleInfo.cpp [CodeGen] Require a name for a block addr target 2019-08-09 20:18:30 +00:00
MachineModuleInfoImpls.cpp Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
MachineOperand.cpp Apply llvm-prefer-register-over-unsigned from clang-tidy to LLVM 2019-08-15 19:22:08 +00:00
MachineOptimizationRemarkEmitter.cpp [llvm] Migrate llvm::make_unique to std::make_unique 2019-08-15 15:54:37 +00:00
MachineOutliner.cpp [Backend] Keep call site info valid through the backend 2019-06-27 13:10:29 +00:00
MachinePipeliner.cpp Apply llvm-prefer-register-over-unsigned from clang-tidy to LLVM 2019-08-15 19:22:08 +00:00
MachinePostDominators.cpp Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
MachineRegionInfo.cpp Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
MachineRegisterInfo.cpp Finish moving TargetRegisterInfo::isVirtualRegister() and friends to llvm::Register as started by r367614. NFC 2019-08-01 23:27:28 +00:00
MachineScheduler.cpp Apply llvm-prefer-register-over-unsigned from clang-tidy to LLVM 2019-08-15 19:22:08 +00:00
MachineSink.cpp [DebugInfo] Make postra sinking of DBG_VALUEs subregister-safe 2019-08-19 09:53:07 +00:00
MachineSSAUpdater.cpp Apply llvm-prefer-register-over-unsigned from clang-tidy to LLVM 2019-08-15 19:22:08 +00:00
MachineTraceMetrics.cpp Apply llvm-prefer-register-over-unsigned from clang-tidy to LLVM 2019-08-15 19:22:08 +00:00
MachineVerifier.cpp [GlobalISel] Introduce a G_DYN_STACKALLOC opcode to represent dynamic allocas. 2019-08-24 02:25:56 +00:00
MacroFusion.cpp [llvm] Migrate llvm::make_unique to std::make_unique 2019-08-15 15:54:37 +00:00
MIRCanonicalizerPass.cpp Apply llvm-prefer-register-over-unsigned from clang-tidy to LLVM 2019-08-15 19:22:08 +00:00
MIRPrinter.cpp [DebugInfo] Allow bundled calls in the MIR's call site info 2019-08-19 12:41:22 +00:00
MIRPrintingPass.cpp Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
OptimizePHIs.cpp Apply llvm-prefer-register-over-unsigned from clang-tidy to LLVM 2019-08-15 19:22:08 +00:00
ParallelCG.cpp Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
PatchableFunction.cpp Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
PeepholeOptimizer.cpp [AMDGPU] Prevent VGPR copies from moving across the EXEC mask definitions 2019-08-21 15:15:04 +00:00
PHIElimination.cpp Apply llvm-prefer-register-over-unsigned from clang-tidy to LLVM 2019-08-15 19:22:08 +00:00
PHIEliminationUtils.cpp Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
PHIEliminationUtils.h Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
PostRAHazardRecognizer.cpp Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
PostRASchedulerList.cpp Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
PreISelIntrinsicLowering.cpp [opaque pointer types] Pass value type to LoadInst creation. 2019-02-01 20:44:24 +00:00
ProcessImplicitDefs.cpp Apply llvm-prefer-register-over-unsigned from clang-tidy to LLVM 2019-08-15 19:22:08 +00:00
PrologEpilogInserter.cpp [LLVM][Alignment] Introduce Alignment In MachineFrameInfo 2019-08-21 14:29:30 +00:00
PseudoSourceValue.cpp [llvm] Migrate llvm::make_unique to std::make_unique 2019-08-15 15:54:37 +00:00
ReachingDefAnalysis.cpp Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
README.txt Test commit: Removed trailing space in .txt file. 2018-12-06 13:20:27 +00:00
RegAllocBase.cpp Finish moving TargetRegisterInfo::isVirtualRegister() and friends to llvm::Register as started by r367614. NFC 2019-08-01 23:27:28 +00:00
RegAllocBase.h Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
RegAllocBasic.cpp Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
RegAllocFast.cpp Apply llvm-prefer-register-over-unsigned from clang-tidy to LLVM 2019-08-15 19:22:08 +00:00
RegAllocGreedy.cpp Apply llvm-prefer-register-over-unsigned from clang-tidy to LLVM 2019-08-15 19:22:08 +00:00
RegAllocPBQP.cpp [llvm] Migrate llvm::make_unique to std::make_unique 2019-08-15 15:54:37 +00:00
RegisterClassInfo.cpp [ARM] Thumb2: favor R4-R7 over R12/LR in allocation order when opt for minsize 2019-07-03 09:58:52 +00:00
RegisterCoalescer.cpp Apply llvm-prefer-register-over-unsigned from clang-tidy to LLVM 2019-08-15 19:22:08 +00:00
RegisterCoalescer.h Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
RegisterPressure.cpp Apply llvm-prefer-register-over-unsigned from clang-tidy to LLVM 2019-08-15 19:22:08 +00:00
RegisterScavenging.cpp RegScavenger: Use Register 2019-08-23 18:25:34 +00:00
RegisterUsageInfo.cpp Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
RegUsageInfoCollector.cpp Reland: Fix and test inter-procedural register allocation for ARM 2019-08-05 09:04:10 +00:00
RegUsageInfoPropagate.cpp [IPRA] Don't rely on non-exact function definitions 2019-07-19 09:59:26 +00:00
RenameIndependentSubregs.cpp Apply llvm-prefer-register-over-unsigned from clang-tidy to LLVM 2019-08-15 19:22:08 +00:00
ResetMachineFunctionPass.cpp [ResetMachineFunctionPass] Add visited functions statistics info 2019-03-14 01:13:15 +00:00
SafeStack.cpp Delete dead stores 2019-07-12 14:58:15 +00:00
SafeStackColoring.cpp Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
SafeStackColoring.h Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
SafeStackLayout.cpp Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
SafeStackLayout.h Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
ScalarizeMaskedMemIntrin.cpp [ScalarizeMaskedMemIntrin] Bitcast the mask to the scalar domain and use scalar bit tests for the branches for expandload/compressstore. 2019-08-02 23:43:53 +00:00
ScheduleDAG.cpp [ScheduleDAGRRList] Recompute topological ordering on demand. 2019-04-17 15:05:29 +00:00
ScheduleDAGInstrs.cpp Apply llvm-prefer-register-over-unsigned from clang-tidy to LLVM 2019-08-15 19:22:08 +00:00
ScheduleDAGPrinter.cpp Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
ScoreboardHazardRecognizer.cpp Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
ShadowStackGCLowering.cpp [opaque pointer types] Pass value type to LoadInst creation. 2019-02-01 20:44:24 +00:00
ShrinkWrap.cpp Apply llvm-prefer-register-over-unsigned from clang-tidy to LLVM 2019-08-15 19:22:08 +00:00
SjLjEHPrepare.cpp Added address-space mangling for stack related intrinsics 2019-07-22 12:42:48 +00:00
SlotIndexes.cpp SlotIndexes: delete unused functions 2019-06-23 16:05:29 +00:00
Spiller.h Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
SpillPlacement.cpp Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
SpillPlacement.h Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
SplitKit.cpp Apply llvm-prefer-register-over-unsigned from clang-tidy to LLVM 2019-08-15 19:22:08 +00:00
SplitKit.h Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
StackColoring.cpp Use llvm::stable_sort 2019-04-23 14:51:27 +00:00
StackMapLivenessAnalysis.cpp Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
StackMaps.cpp Apply llvm-prefer-register-over-unsigned from clang-tidy to LLVM 2019-08-15 19:22:08 +00:00
StackProtector.cpp StackProtector: Use PointerMayBeCaptured 2019-06-12 14:23:33 +00:00
StackSlotColoring.cpp Finish moving TargetRegisterInfo::isVirtualRegister() and friends to llvm::Register as started by r367614. NFC 2019-08-01 23:27:28 +00:00
SwiftErrorValueTracking.cpp GlobalISel: Remove unsigned variant of SrcOp 2019-06-24 16:16:12 +00:00
SwitchLoweringUtils.cpp [GlobalISel][IRTranslator] Change switch table translation to generate jump tables and range checks. 2019-06-21 18:10:38 +00:00
TailDuplication.cpp Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
TailDuplicator.cpp Apply llvm-prefer-register-over-unsigned from clang-tidy to LLVM 2019-08-15 19:22:08 +00:00
TargetFrameLoweringImpl.cpp [IPRA][ARM] Disable no-CSR optimisation for ARM 2019-08-02 10:23:17 +00:00
TargetInstrInfo.cpp Apply llvm-prefer-register-over-unsigned from clang-tidy to LLVM 2019-08-15 19:22:08 +00:00
TargetLoweringBase.cpp [LLVM][NFC] remove unused fields 2019-08-23 20:49:06 +00:00
TargetLoweringObjectFileImpl.cpp [PowerPC][AIX] Adds support for writing the .data section in assembly files 2019-08-25 15:17:25 +00:00
TargetOptionsImpl.cpp Update the file headers across all of the LLVM projects in the monorepo 2019-01-19 08:50:56 +00:00
TargetPassConfig.cpp Do a sweep of symbol internalization. NFC. 2019-08-23 19:59:23 +00:00
TargetRegisterInfo.cpp Eliminate implicit Register->unsigned conversions in VirtRegMap. NFC 2019-08-13 00:55:24 +00:00
TargetSchedule.cpp Apply llvm-prefer-register-over-unsigned from clang-tidy to LLVM 2019-08-15 19:22:08 +00:00
TargetSubtargetInfo.cpp [Subtarget] Merge ProcSched and ProcDesc arrays in MCSubtargetInfo into a single array. 2019-03-05 18:54:38 +00:00
TwoAddressInstructionPass.cpp Apply llvm-prefer-register-over-unsigned from clang-tidy to LLVM 2019-08-15 19:22:08 +00:00
UnreachableBlockElim.cpp Apply llvm-prefer-register-over-unsigned from clang-tidy to LLVM 2019-08-15 19:22:08 +00:00
ValueTypes.cpp [MVT] Add v16f16 and v32f16 vectors. 2019-08-21 19:14:48 +00:00
VirtRegMap.cpp Eliminate implicit Register->unsigned conversions in VirtRegMap. NFC 2019-08-13 00:55:24 +00:00
WasmEHPrepare.cpp [WebAssembly] Make rethrow take an except_ref type argument 2019-03-16 05:38:57 +00:00
WinEHPrepare.cpp Fix parameter name comments using clang-tidy. NFC. 2019-07-16 04:46:31 +00:00
XRayInstrumentation.cpp [Backend] Keep call site info valid through the backend 2019-06-27 13:10:29 +00:00

//===---------------------------------------------------------------------===//

Common register allocation / spilling problem:

        mul lr, r4, lr
        str lr, [sp, #+52]
        ldr lr, [r1, #+32]
        sxth r3, r3
        ldr r4, [sp, #+52]
        mla r4, r3, lr, r4

can be:

        mul lr, r4, lr
        mov r4, lr
        str lr, [sp, #+52]
        ldr lr, [r1, #+32]
        sxth r3, r3
        mla r4, r3, lr, r4

and then "merge" mul and mov:

        mul r4, r4, lr
        str r4, [sp, #+52]
        ldr lr, [r1, #+32]
        sxth r3, r3
        mla r4, r3, lr, r4

It also increase the likelihood the store may become dead.

//===---------------------------------------------------------------------===//

bb27 ...
        ...
        %reg1037 = ADDri %reg1039, 1
        %reg1038 = ADDrs %reg1032, %reg1039, %noreg, 10
    Successors according to CFG: 0x8b03bf0 (#5)

bb76 (0x8b03bf0, LLVM BB @0x8b032d0, ID#5):
    Predecessors according to CFG: 0x8b0c5f0 (#3) 0x8b0a7c0 (#4)
        %reg1039 = PHI %reg1070, mbb<bb76.outer,0x8b0c5f0>, %reg1037, mbb<bb27,0x8b0a7c0>

Note ADDri is not a two-address instruction. However, its result %reg1037 is an
operand of the PHI node in bb76 and its operand %reg1039 is the result of the
PHI node. We should treat it as a two-address code and make sure the ADDri is
scheduled after any node that reads %reg1039.

//===---------------------------------------------------------------------===//

Use local info (i.e. register scavenger) to assign it a free register to allow
reuse:
        ldr r3, [sp, #+4]
        add r3, r3, #3
        ldr r2, [sp, #+8]
        add r2, r2, #2
        ldr r1, [sp, #+4]  <==
        add r1, r1, #1
        ldr r0, [sp, #+4]
        add r0, r0, #2

//===---------------------------------------------------------------------===//

LLVM aggressively lift CSE out of loop. Sometimes this can be negative side-
effects:

R1 = X + 4
R2 = X + 7
R3 = X + 15

loop:
load [i + R1]
...
load [i + R2]
...
load [i + R3]

Suppose there is high register pressure, R1, R2, R3, can be spilled. We need
to implement proper re-materialization to handle this:

R1 = X + 4
R2 = X + 7
R3 = X + 15

loop:
R1 = X + 4  @ re-materialized
load [i + R1]
...
R2 = X + 7 @ re-materialized
load [i + R2]
...
R3 = X + 15 @ re-materialized
load [i + R3]

Furthermore, with re-association, we can enable sharing:

R1 = X + 4
R2 = X + 7
R3 = X + 15

loop:
T = i + X
load [T + 4]
...
load [T + 7]
...
load [T + 15]
//===---------------------------------------------------------------------===//

It's not always a good idea to choose rematerialization over spilling. If all
the load / store instructions would be folded then spilling is cheaper because
it won't require new live intervals / registers. See 2003-05-31-LongShifts for
an example.

//===---------------------------------------------------------------------===//

With a copying garbage collector, derived pointers must not be retained across
collector safe points; the collector could move the objects and invalidate the
derived pointer. This is bad enough in the first place, but safe points can
crop up unpredictably. Consider:

        %array = load { i32, [0 x %obj] }** %array_addr
        %nth_el = getelementptr { i32, [0 x %obj] }* %array, i32 0, i32 %n
        %old = load %obj** %nth_el
        %z = div i64 %x, %y
        store %obj* %new, %obj** %nth_el

If the i64 division is lowered to a libcall, then a safe point will (must)
appear for the call site. If a collection occurs, %array and %nth_el no longer
point into the correct object.

The fix for this is to copy address calculations so that dependent pointers
are never live across safe point boundaries. But the loads cannot be copied
like this if there was an intervening store, so may be hard to get right.

Only a concurrent mutator can trigger a collection at the libcall safe point.
So single-threaded programs do not have this requirement, even with a copying
collector. Still, LLVM optimizations would probably undo a front-end's careful
work.

//===---------------------------------------------------------------------===//

The ocaml frametable structure supports liveness information. It would be good
to support it.

//===---------------------------------------------------------------------===//

The FIXME in ComputeCommonTailLength in BranchFolding.cpp needs to be
revisited. The check is there to work around a misuse of directives in inline
assembly.

//===---------------------------------------------------------------------===//

It would be good to detect collector/target compatibility instead of silently
doing the wrong thing.

//===---------------------------------------------------------------------===//

It would be really nice to be able to write patterns in .td files for copies,
which would eliminate a bunch of explicit predicates on them (e.g. no side
effects).  Once this is in place, it would be even better to have tblgen
synthesize the various copy insertion/inspection methods in TargetInstrInfo.

//===---------------------------------------------------------------------===//

Stack coloring improvements:

1. Do proper LiveStacks analysis on all stack objects including those which are
   not spill slots.
2. Reorder objects to fill in gaps between objects.
   e.g. 4, 1, <gap>, 4, 1, 1, 1, <gap>, 4 => 4, 1, 1, 1, 1, 4, 4

//===---------------------------------------------------------------------===//

The scheduler should be able to sort nearby instructions by their address. For
example, in an expanded memset sequence it's not uncommon to see code like this:

  movl $0, 4(%rdi)
  movl $0, 8(%rdi)
  movl $0, 12(%rdi)
  movl $0, 0(%rdi)

Each of the stores is independent, and the scheduler is currently making an
arbitrary decision about the order.

//===---------------------------------------------------------------------===//

Another opportunitiy in this code is that the $0 could be moved to a register:

  movl $0, 4(%rdi)
  movl $0, 8(%rdi)
  movl $0, 12(%rdi)
  movl $0, 0(%rdi)

This would save substantial code size, especially for longer sequences like
this. It would be easy to have a rule telling isel to avoid matching MOV32mi
if the immediate has more than some fixed number of uses. It's more involved
to teach the register allocator how to do late folding to recover from
excessive register pressure.