1
0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-11-25 04:02:41 +01:00
llvm-mirror/lib/CodeGen
2009-11-05 01:16:59 +00:00
..
AsmPrinter While calculating original type size for a derived type, handle type variants encoded as DIDerivedType appropriately. 2009-11-04 23:48:00 +00:00
PBQP
SelectionDAG Make -print-machineinstrs more readable. 2009-10-31 20:19:03 +00:00
AggressiveAntiDepBreaker.cpp Do a scheduling pass ignoring anti-dependencies to identify candidate registers that should be renamed. 2009-11-03 20:57:50 +00:00
AggressiveAntiDepBreaker.h Do a scheduling pass ignoring anti-dependencies to identify candidate registers that should be renamed. 2009-11-03 20:57:50 +00:00
AntiDepBreaker.h Do a scheduling pass ignoring anti-dependencies to identify candidate registers that should be renamed. 2009-11-03 20:57:50 +00:00
BranchFolding.cpp Fix branch folding bug for indirect branches: for a block containing only 2009-11-03 23:44:31 +00:00
BranchFolding.h Revert r85346 change to control tail merging by CodeGenOpt::Level. 2009-10-28 20:46:46 +00:00
CMakeLists.txt Fix CMake makefiles 2009-11-04 01:32:06 +00:00
CodePlacementOpt.cpp
CriticalAntiDepBreaker.cpp Do a scheduling pass ignoring anti-dependencies to identify candidate registers that should be renamed. 2009-11-03 20:57:50 +00:00
CriticalAntiDepBreaker.h Do a scheduling pass ignoring anti-dependencies to identify candidate registers that should be renamed. 2009-11-03 20:57:50 +00:00
DeadMachineInstructionElim.cpp
DwarfEHPrepare.cpp Don't put in these EH changes. 2009-10-29 00:37:35 +00:00
ELF.h
ELFCodeEmitter.cpp
ELFCodeEmitter.h
ELFWriter.cpp
ELFWriter.h
ExactHazardRecognizer.cpp Do a scheduling pass ignoring anti-dependencies to identify candidate registers that should be renamed. 2009-11-03 20:57:50 +00:00
ExactHazardRecognizer.h
GCMetadata.cpp
GCMetadataPrinter.cpp
GCStrategy.cpp
IfConversion.cpp Revert r85346 change to control tail merging by CodeGenOpt::Level. 2009-10-28 20:46:46 +00:00
IntrinsicLowering.cpp
LatencyPriorityQueue.cpp Do a scheduling pass ignoring anti-dependencies to identify candidate registers that should be renamed. 2009-11-03 20:57:50 +00:00
LiveInterval.cpp The Indexes Patch. 2009-11-03 23:52:08 +00:00
LiveIntervalAnalysis.cpp The Indexes Patch. 2009-11-03 23:52:08 +00:00
LiveStackAnalysis.cpp The Indexes Patch. 2009-11-03 23:52:08 +00:00
LiveVariables.cpp
LLVMTargetMachine.cpp Now that code placement optimization pass is run for JIT, make sure it's before pre-emit passes. 2009-11-05 01:16:59 +00:00
LowerSubregs.cpp
MachineBasicBlock.cpp Make -print-machineinstrs more readable. 2009-10-31 20:19:03 +00:00
MachineDominators.cpp
MachineFunction.cpp Make -print-machineinstrs more readable. 2009-10-31 20:19:03 +00:00
MachineFunctionAnalysis.cpp
MachineFunctionPass.cpp
MachineInstr.cpp Make -print-machineinstrs more readable. 2009-10-31 20:19:03 +00:00
MachineLICM.cpp Code refactoring. 2009-11-05 00:51:13 +00:00
MachineLoopInfo.cpp
MachineModuleInfo.cpp
MachineModuleInfoImpls.cpp
MachinePassRegistry.cpp
MachineRegisterInfo.cpp
MachineSink.cpp
MachineVerifier.cpp Make -print-machineinstrs more readable. 2009-10-31 20:19:03 +00:00
MachO.h
MachOCodeEmitter.cpp
MachOCodeEmitter.h
MachOWriter.cpp
MachOWriter.h
Makefile
ObjectCodeEmitter.cpp
OcamlGC.cpp
Passes.cpp
PHIElimination.cpp
PHIElimination.h
PostRASchedulerList.cpp Do a scheduling pass ignoring anti-dependencies to identify candidate registers that should be renamed. 2009-11-03 20:57:50 +00:00
PreAllocSplitting.cpp The Indexes Patch. 2009-11-03 23:52:08 +00:00
ProcessImplicitDefs.cpp The Indexes Patch. 2009-11-03 23:52:08 +00:00
PrologEpilogInserter.cpp When the function is doing dynamic stack realignment, the spill slot will be 2009-10-29 02:33:47 +00:00
PrologEpilogInserter.h
PseudoSourceValue.cpp Add PseudoSourceValue::mayAlias. It returns true if the object can ever alias any LLVM IR value. 2009-11-01 23:50:04 +00:00
README.txt
RegAllocLinearScan.cpp The Indexes Patch. 2009-11-03 23:52:08 +00:00
RegAllocLocal.cpp
RegAllocPBQP.cpp The Indexes Patch. 2009-11-03 23:52:08 +00:00
RegisterCoalescer.cpp
RegisterScavenging.cpp
ScheduleDAG.cpp Do a scheduling pass ignoring anti-dependencies to identify candidate registers that should be renamed. 2009-11-03 20:57:50 +00:00
ScheduleDAGEmit.cpp Add assertion checks here to turn silent miscompiles into aborts. 2009-10-30 23:59:06 +00:00
ScheduleDAGInstrs.cpp Correctly add chain dependencies around calls and unknown-side-effect instructions. 2009-11-05 00:16:44 +00:00
ScheduleDAGInstrs.h
ScheduleDAGPrinter.cpp
ShadowStackGC.cpp
ShrinkWrapping.cpp
SimpleHazardRecognizer.h
SimpleRegisterCoalescing.cpp RangeIsDefinedByCopyFromReg() should check for subreg_to_reg, insert_subreg, 2009-11-04 08:33:14 +00:00
SimpleRegisterCoalescing.h The Indexes Patch. 2009-11-03 23:52:08 +00:00
SjLjEHPrepare.cpp
SlotIndexes.cpp Handle empty/tombstone keys for LiveIndex more cleanly. Check for index sanity when constructing index list entries. 2009-11-04 21:24:15 +00:00
Spiller.cpp The Indexes Patch. 2009-11-03 23:52:08 +00:00
Spiller.h
StackProtector.cpp
StackSlotColoring.cpp The Indexes Patch. 2009-11-03 23:52:08 +00:00
StrongPHIElimination.cpp The Indexes Patch. 2009-11-03 23:52:08 +00:00
TargetInstrInfoImpl.cpp
TwoAddressInstructionPass.cpp
UnreachableBlockElim.cpp
VirtRegMap.cpp The Indexes Patch. 2009-11-03 23:52:08 +00:00
VirtRegMap.h The Indexes Patch. 2009-11-03 23:52:08 +00:00
VirtRegRewriter.cpp When there is a 2-instruction spill sequence, record 2009-10-29 01:15:40 +00:00
VirtRegRewriter.h

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

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

It also increase the likelyhood 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 improvments:

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.