llvm-6502/lib/CodeGen
Oliver Stannard 5e487f8dc7 Teach the AArch64 backend about v4f16 and v8f16
This teaches the AArch64 backend to deal with the operations required
to deal with the operations on v4f16 and v8f16 which are exposed by
NEON intrinsics, plus the add, sub, mul and div operations.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@216555 91177308-0d34-0410-b5e6-96231b3b80d8
2014-08-27 16:16:04 +00:00
..
AsmPrinter Fix some cases were ArrayRefs were being passed by reference. Also remove 'const' from some other ArrayRef uses since its implicitly const already. 2014-08-27 05:25:00 +00:00
SelectionDAG Teach the AArch64 backend about v4f16 and v8f16 2014-08-27 16:16:04 +00:00
AggressiveAntiDepBreaker.cpp Have MachineFunction cache a pointer to the subtarget to make lookups 2014-08-05 02:39:49 +00:00
AggressiveAntiDepBreaker.h Canonicalize header guards into a common format. 2014-08-13 16:26:38 +00:00
AllocationOrder.cpp
AllocationOrder.h Canonicalize header guards into a common format. 2014-08-13 16:26:38 +00:00
Analysis.cpp Remove the TargetMachine forwards for TargetSubtargetInfo based 2014-08-04 21:25:23 +00:00
AntiDepBreaker.h Canonicalize header guards into a common format. 2014-08-13 16:26:38 +00:00
AtomicExpandPass.cpp Rename AtomicExpandLoadLinked into AtomicExpand 2014-08-21 21:50:01 +00:00
BasicTargetTransformInfo.cpp Allow vectorization of division by uniform power of 2. 2014-08-25 04:56:54 +00:00
BranchFolding.cpp [Branch probability] Recompute branch weights of tail-merged basic blocks. 2014-08-07 19:30:13 +00:00
BranchFolding.h Canonicalize header guards into a common format. 2014-08-13 16:26:38 +00:00
CalcSpillWeights.cpp Have MachineFunction cache a pointer to the subtarget to make lookups 2014-08-05 02:39:49 +00:00
CallingConvLower.cpp Remove the target machine from CCState. Previously it was only used 2014-08-06 18:45:26 +00:00
CMakeLists.txt Rename AtomicExpandLoadLinked into AtomicExpand 2014-08-21 21:50:01 +00:00
CodeGen.cpp Rename AtomicExpandLoadLinked into AtomicExpand 2014-08-21 21:50:01 +00:00
CodeGenPrepare.cpp Repace SmallPtrSet with SmallPtrSetImpl in function arguments to avoid needing to mention the size. 2014-08-21 05:55:13 +00:00
CriticalAntiDepBreaker.cpp critical-anti-dependency breaker: don't use reg def info from kill insts (PR20308) 2014-08-20 18:03:00 +00:00
CriticalAntiDepBreaker.h Canonicalize header guards into a common format. 2014-08-13 16:26:38 +00:00
DeadMachineInstructionElim.cpp [Cleanup] Utility function to erase instruction and mark DBG_Values 2014-08-13 21:15:23 +00:00
DFAPacketizer.cpp Cleanup: Delete seemingly unused reference to MachineDominatorTree from ScheduleDAGInstrs. 2014-08-20 20:57:26 +00:00
DwarfEHPrepare.cpp Remove the TargetMachine forwards for TargetSubtargetInfo based 2014-08-04 21:25:23 +00:00
EarlyIfConversion.cpp Have MachineFunction cache a pointer to the subtarget to make lookups 2014-08-05 02:39:49 +00:00
EdgeBundles.cpp
ErlangGC.cpp Have MachineFunction cache a pointer to the subtarget to make lookups 2014-08-05 02:39:49 +00:00
ExecutionDepsFix.cpp Have MachineFunction cache a pointer to the subtarget to make lookups 2014-08-05 02:39:49 +00:00
ExpandISelPseudos.cpp Have MachineFunction cache a pointer to the subtarget to make lookups 2014-08-05 02:39:49 +00:00
ExpandPostRAPseudos.cpp Have MachineFunction cache a pointer to the subtarget to make lookups 2014-08-05 02:39:49 +00:00
GCMetadata.cpp
GCMetadataPrinter.cpp
GCStrategy.cpp Remove the TargetMachine forwards for TargetSubtargetInfo based 2014-08-04 21:25:23 +00:00
GlobalMerge.cpp Remove the TargetMachine forwards for TargetSubtargetInfo based 2014-08-04 21:25:23 +00:00
IfConversion.cpp CodeGen: switch to a range based for loop 2014-08-09 17:21:29 +00:00
InlineSpiller.cpp Have MachineFunction cache a pointer to the subtarget to make lookups 2014-08-05 02:39:49 +00:00
InterferenceCache.cpp
InterferenceCache.h Canonicalize header guards into a common format. 2014-08-13 16:26:38 +00:00
IntrinsicLowering.cpp Add missing Interpreter intrinsic lowering for sin, cos and ceil 2014-08-08 15:00:12 +00:00
JITCodeEmitter.cpp Temporarily Revert "Nuke the old JIT." as it's not quite ready to 2014-08-07 22:02:54 +00:00
JumpInstrTables.cpp Don't manually (and forcibly) run the verifier on the entire module from 2014-07-30 05:44:04 +00:00
LatencyPriorityQueue.cpp
LexicalScopes.cpp Repace SmallPtrSet with SmallPtrSetImpl in function arguments to avoid needing to mention the size. 2014-08-21 05:55:13 +00:00
LiveDebugVariables.cpp test commit: remove trailing whitespace. 2014-08-07 20:04:00 +00:00
LiveDebugVariables.h Canonicalize header guards into a common format. 2014-08-13 16:26:38 +00:00
LiveInterval.cpp
LiveIntervalAnalysis.cpp Remove the TargetMachine forwards for TargetSubtargetInfo based 2014-08-04 21:25:23 +00:00
LiveIntervalUnion.cpp
LivePhysRegs.cpp
LiveRangeCalc.cpp
LiveRangeCalc.h Canonicalize header guards into a common format. 2014-08-13 16:26:38 +00:00
LiveRangeEdit.cpp Add TargetInstrInfo interface isAsCheapAsAMove. 2014-07-29 01:55:19 +00:00
LiveRegMatrix.cpp Have MachineFunction cache a pointer to the subtarget to make lookups 2014-08-05 02:39:49 +00:00
LiveStackAnalysis.cpp Have MachineFunction cache a pointer to the subtarget to make lookups 2014-08-05 02:39:49 +00:00
LiveVariables.cpp CodeGen/LiveVariables: use vector::assign() 2014-08-26 02:03:25 +00:00
LLVMBuild.txt
LLVMTargetMachine.cpp Temporarily Revert "Nuke the old JIT." as it's not quite ready to 2014-08-07 22:02:54 +00:00
LocalStackSlotAllocation.cpp Have MachineFunction cache a pointer to the subtarget to make lookups 2014-08-05 02:39:49 +00:00
MachineBasicBlock.cpp [MachineDominatorTree] Provide a method to inform a MachineDominatorTree that a 2014-08-13 21:00:07 +00:00
MachineBlockFrequencyInfo.cpp Revert "Introduce a string_ostream string builder facilty" 2014-06-26 22:52:05 +00:00
MachineBlockPlacement.cpp Have MachineFunction cache a pointer to the subtarget to make lookups 2014-08-05 02:39:49 +00:00
MachineBranchProbabilityInfo.cpp
MachineCodeEmitter.cpp
MachineCombiner.cpp [MachineCombiner] Removal of dangling DBG_VALUES after combining [20598] 2014-08-13 22:07:36 +00:00
MachineCopyPropagation.cpp Have MachineFunction cache a pointer to the subtarget to make lookups 2014-08-05 02:39:49 +00:00
MachineCSE.cpp In Machine CSE pass, the source register of a COPY machine instruction can 2014-08-11 05:17:19 +00:00
MachineDominanceFrontier.cpp Templatify DominanceFrontier. 2014-07-12 21:59:52 +00:00
MachineDominators.cpp [MachineDominatorTree] Provide a method to inform a MachineDominatorTree that a 2014-08-13 21:00:07 +00:00
MachineFunction.cpp Make isAliased property for fixed-offset stack objects adjustable 2014-08-16 00:17:02 +00:00
MachineFunctionAnalysis.cpp
MachineFunctionPass.cpp
MachineFunctionPrinterPass.cpp
MachineInstr.cpp [Cleanup] Utility function to erase instruction and mark DBG_Values 2014-08-13 21:15:23 +00:00
MachineInstrBundle.cpp Remove the TargetMachine forwards for TargetSubtargetInfo based 2014-08-04 21:25:23 +00:00
MachineLICM.cpp Remove the TargetMachine forwards for TargetSubtargetInfo based 2014-08-04 21:25:23 +00:00
MachineLoopInfo.cpp
MachineModuleInfo.cpp Delete dead code. NFC. 2014-08-15 14:58:22 +00:00
MachineModuleInfoImpls.cpp
MachinePassRegistry.cpp
MachinePostDominators.cpp
MachineRegionInfo.cpp Header hygiene: remove using directive and #undef DEBUG_TYPE once we're done. 2014-07-30 00:25:24 +00:00
MachineRegisterInfo.cpp CodeGen: switch raw array to std::vector 2014-08-25 00:28:31 +00:00
MachineScheduler.cpp Debugging Utility - optional ability for dumping critical path length 2014-08-07 21:49:44 +00:00
MachineSink.cpp [MachineSink] Improve the compile time by preserving the dominance information 2014-08-11 23:52:01 +00:00
MachineSSAUpdater.cpp Have MachineFunction cache a pointer to the subtarget to make lookups 2014-08-05 02:39:49 +00:00
MachineTraceMetrics.cpp Simplify creation of a bunch of ArrayRefs by using None, makeArrayRef or just letting them be implicitly created. 2014-08-27 05:25:25 +00:00
MachineVerifier.cpp Modernize raw_fd_ostream's constructor a bit. 2014-08-25 18:16:47 +00:00
Makefile
module.modulemap
OcamlGC.cpp
OptimizePHIs.cpp Have MachineFunction cache a pointer to the subtarget to make lookups 2014-08-05 02:39:49 +00:00
Passes.cpp [PeepholeOptimizer] Refactor the advanced copy optimization to take advantage of 2014-08-20 17:41:48 +00:00
PeepholeOptimizer.cpp [PeepholeOptimizer] Enable the advanced copy optimization by default. 2014-08-21 22:23:52 +00:00
PHIElimination.cpp Use range based for loops to avoid needing to re-mention SmallPtrSet size. 2014-08-24 23:23:06 +00:00
PHIEliminationUtils.cpp
PHIEliminationUtils.h Canonicalize header guards into a common format. 2014-08-13 16:26:38 +00:00
PostRASchedulerList.cpp Cleanup: Delete seemingly unused reference to MachineDominatorTree from ScheduleDAGInstrs. 2014-08-20 20:57:26 +00:00
ProcessImplicitDefs.cpp Have MachineFunction cache a pointer to the subtarget to make lookups 2014-08-05 02:39:49 +00:00
PrologEpilogInserter.cpp Use range based for loops to avoid needing to re-mention SmallPtrSet size. 2014-08-24 23:23:06 +00:00
PrologEpilogInserter.h Canonicalize header guards into a common format. 2014-08-13 16:26:38 +00:00
PseudoSourceValue.cpp Make isAliased property for fixed-offset stack objects adjustable 2014-08-16 00:17:02 +00:00
README.txt
RegAllocBase.cpp
RegAllocBase.h Canonicalize header guards into a common format. 2014-08-13 16:26:38 +00:00
RegAllocBasic.cpp Remove uses of the redundant ".reset(nullptr)" of unique_ptr, in favor of ".reset()" 2014-07-19 01:05:11 +00:00
RegAllocFast.cpp Use range based for loops to avoid needing to re-mention SmallPtrSet size. 2014-08-24 23:23:06 +00:00
RegAllocGreedy.cpp Simplify creation of a bunch of ArrayRefs by using None, makeArrayRef or just letting them be implicitly created. 2014-08-27 05:25:25 +00:00
RegAllocPBQP.cpp Modernize raw_fd_ostream's constructor a bit. 2014-08-25 18:16:47 +00:00
RegisterClassInfo.cpp Have MachineFunction cache a pointer to the subtarget to make lookups 2014-08-05 02:39:49 +00:00
RegisterCoalescer.cpp Repace SmallPtrSet with SmallPtrSetImpl in function arguments to avoid needing to mention the size. 2014-08-21 05:55:13 +00:00
RegisterCoalescer.h Canonicalize header guards into a common format. 2014-08-13 16:26:38 +00:00
RegisterPressure.cpp Have MachineFunction cache a pointer to the subtarget to make lookups 2014-08-05 02:39:49 +00:00
RegisterScavenging.cpp Changed the liveness tracking in the RegisterScavenger 2014-08-04 23:07:49 +00:00
ScheduleDAG.cpp Remove the TargetMachine forwards for TargetSubtargetInfo based 2014-08-04 21:25:23 +00:00
ScheduleDAGInstrs.cpp Repace SmallPtrSet with SmallPtrSetImpl in function arguments to avoid needing to mention the size. 2014-08-21 05:55:13 +00:00
ScheduleDAGPrinter.cpp
ScoreboardHazardRecognizer.cpp
ShadowStackGC.cpp
SjLjEHPrepare.cpp Use range based for loops to avoid needing to re-mention SmallPtrSet size. 2014-08-24 23:23:06 +00:00
SlotIndexes.cpp
Spiller.cpp Have MachineFunction cache a pointer to the subtarget to make lookups 2014-08-05 02:39:49 +00:00
Spiller.h Canonicalize header guards into a common format. 2014-08-13 16:26:38 +00:00
SpillPlacement.cpp
SpillPlacement.h Canonicalize header guards into a common format. 2014-08-13 16:26:38 +00:00
SplitKit.cpp Have MachineFunction cache a pointer to the subtarget to make lookups 2014-08-05 02:39:49 +00:00
SplitKit.h Canonicalize header guards into a common format. 2014-08-13 16:26:38 +00:00
StackColoring.cpp
StackMapLivenessAnalysis.cpp Have MachineFunction cache a pointer to the subtarget to make lookups 2014-08-05 02:39:49 +00:00
StackMaps.cpp Have MachineFunction cache a pointer to the subtarget to make lookups 2014-08-05 02:39:49 +00:00
StackProtector.cpp Remove the TargetMachine forwards for TargetSubtargetInfo based 2014-08-04 21:25:23 +00:00
StackSlotColoring.cpp Have MachineFunction cache a pointer to the subtarget to make lookups 2014-08-05 02:39:49 +00:00
TailDuplication.cpp Have MachineFunction cache a pointer to the subtarget to make lookups 2014-08-05 02:39:49 +00:00
TargetFrameLoweringImpl.cpp Have MachineFunction cache a pointer to the subtarget to make lookups 2014-08-05 02:39:49 +00:00
TargetInstrInfo.cpp Add isInsertSubreg property. 2014-08-20 23:49:36 +00:00
TargetLoweringBase.cpp name change: isPow2DivCheap -> isPow2SDivCheap 2014-08-21 22:31:48 +00:00
TargetLoweringObjectFileImpl.cpp Remove the TargetMachine forwards for TargetSubtargetInfo based 2014-08-04 21:25:23 +00:00
TargetOptionsImpl.cpp
TargetRegisterInfo.cpp
TargetSchedule.cpp Have MachineFunction cache a pointer to the subtarget to make lookups 2014-08-05 02:39:49 +00:00
TwoAddressInstructionPass.cpp Remove the TargetMachine forwards for TargetSubtargetInfo based 2014-08-04 21:25:23 +00:00
UnreachableBlockElim.cpp Use range based for loops to avoid needing to re-mention SmallPtrSet size. 2014-08-24 23:23:06 +00:00
VirtRegMap.cpp Have MachineFunction cache a pointer to the subtarget to make lookups 2014-08-05 02:39:49 +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 LiveStackAnalysis 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.