llvm-project/llvm/lib/CodeGen
Owen Anderson 97364655dc Significant improvements to the logic for merging live intervals. This code can't
just use LI::MergeValueAsValue, as its behavior in the presence of overlapping ranges
isn't what StrongPHIElimination wants.

llvm-svn: 56472
2008-09-22 21:58:58 +00:00
..
AsmPrinter Initial support for the CMake build system. 2008-09-22 01:08:49 +00:00
SelectionDAG Make log, log2, log10, exp, exp2 use Expand by 2008-09-22 21:57:32 +00:00
BranchFolding.cpp Mark several codegen passes as preserving all analysis. 2008-09-22 20:58:04 +00:00
CMakeLists.txt Initial support for the CMake build system. 2008-09-22 01:08:49 +00:00
DeadMachineInstructionElim.cpp Don't consider instructions with implicit physical register 2008-09-18 18:22:32 +00:00
ELFWriter.cpp Tidy up several unbeseeming casts from pointer to intptr_t. 2008-09-04 17:05:41 +00:00
ELFWriter.h Use raw_ostream throughout the AsmPrinter. 2008-08-21 00:14:44 +00:00
GCMetadata.cpp Tidy up several unbeseeming casts from pointer to intptr_t. 2008-09-04 17:05:41 +00:00
GCMetadataPrinter.cpp Use raw_ostream throughout the AsmPrinter. 2008-08-21 00:14:44 +00:00
GCStrategy.cpp Tidy up several unbeseeming casts from pointer to intptr_t. 2008-09-04 17:05:41 +00:00
IfConversion.cpp Tidy up several unbeseeming casts from pointer to intptr_t. 2008-09-04 17:05:41 +00:00
IntrinsicLowering.cpp More refactoring. Yawn. 2008-09-22 20:51:30 +00:00
LLVMTargetMachine.cpp Use raw_ostream throughout the AsmPrinter. 2008-08-21 00:14:44 +00:00
LiveInterval.cpp Remove AsmThatEarlyClobber etc. from LiveIntervalAnalysis 2008-09-19 01:02:35 +00:00
LiveIntervalAnalysis.cpp Fetch the starting index of the block when assigning intervals. This gets live-in indices 2008-09-21 20:43:24 +00:00
LiveStackAnalysis.cpp Mark several codegen passes as preserving all analysis. 2008-09-22 20:58:04 +00:00
LiveVariables.cpp Factor out code into HandleVirtRegDef, for consistency with 2008-09-21 21:11:41 +00:00
LoopAligner.cpp Mark several codegen passes as preserving all analysis. 2008-09-22 20:58:04 +00:00
LowerSubregs.cpp Mark several codegen passes as preserving all analysis. 2008-09-22 20:58:04 +00:00
MachOWriter.cpp Tidy up several unbeseeming casts from pointer to intptr_t. 2008-09-04 17:05:41 +00:00
MachOWriter.h Use raw_ostream throughout the AsmPrinter. 2008-08-21 00:14:44 +00:00
MachineBasicBlock.cpp Remove isImm(), isReg(), and friends, in favor of 2008-09-13 17:58:21 +00:00
MachineDominators.cpp Change class' public PassInfo variables to by initialized with the 2008-05-13 02:05:11 +00:00
MachineFunction.cpp Fix these comments to reflect current reality. Surprisingly, 2008-09-16 20:45:53 +00:00
MachineInstr.cpp Add a bit to mark operands of asm's that conflict 2008-09-17 21:13:11 +00:00
MachineLICM.cpp Tidy up several unbeseeming casts from pointer to intptr_t. 2008-09-04 17:05:41 +00:00
MachineLoopInfo.cpp Change class' public PassInfo variables to by initialized with the 2008-05-13 02:05:11 +00:00
MachineModuleInfo.cpp Mark several codegen passes as preserving all analysis. 2008-09-22 20:58:04 +00:00
MachinePassRegistry.cpp Remove attribution from file headers, per discussion on llvmdev. 2007-12-29 20:36:04 +00:00
MachineRegisterInfo.cpp Assert that all MachineInstrs update PhysRegUseDefLists in 2008-07-07 19:55:35 +00:00
MachineSink.cpp Remove isImm(), isReg(), and friends, in favor of 2008-09-13 17:58:21 +00:00
Makefile Move all assembler printing related stuff into new libAsmPrinter 2008-08-17 13:53:04 +00:00
OcamlGC.cpp Rename some GC classes so that their roll will hopefully be clearer. 2008-08-17 18:44:35 +00:00
PHIElimination.cpp Tidy up several unbeseeming casts from pointer to intptr_t. 2008-09-04 17:05:41 +00:00
Passes.cpp Clean up the use of static and anonymous namespaces. This turned up 2008-05-13 00:00:25 +00:00
PhysRegTracker.h Rename MRegisterInfo to TargetRegisterInfo. 2008-02-10 18:45:23 +00:00
PostRASchedulerList.cpp Tidy up several unbeseeming casts from pointer to intptr_t. 2008-09-04 17:05:41 +00:00
PrologEpilogInserter.cpp Mark several codegen passes as preserving all analysis. 2008-09-22 20:58:04 +00:00
PseudoSourceValue.cpp Remove the std::ostream form of PseudoSourceValue's print, 2008-08-27 16:19:44 +00:00
README.txt Remove tabs. 2008-08-22 00:04:26 +00:00
RegAllocBigBlock.cpp Tidy up several unbeseeming casts from pointer to intptr_t. 2008-09-04 17:05:41 +00:00
RegAllocLinearScan.cpp Teach coalescer about earlyclobber bits. 2008-09-20 02:03:04 +00:00
RegAllocLocal.cpp Remove isImm(), isReg(), and friends, in favor of 2008-09-13 17:58:21 +00:00
RegAllocSimple.cpp Tidy up several unbeseeming casts from pointer to intptr_t. 2008-09-04 17:05:41 +00:00
RegisterCoalescer.cpp Clean up the use of static and anonymous namespaces. This turned up 2008-05-13 00:00:25 +00:00
RegisterScavenging.cpp Minor const-correctness fixes. 2008-07-07 20:06:06 +00:00
ShadowStackGC.cpp Rename some GC classes so that their roll will hopefully be clearer. 2008-08-17 18:44:35 +00:00
SimpleRegisterCoalescing.cpp Mark several codegen passes as preserving all analysis. 2008-09-22 20:58:04 +00:00
SimpleRegisterCoalescing.h Re-materalized definition instructions may be dead. Whack them. 2008-09-19 17:38:47 +00:00
StackSlotColoring.cpp Mark several codegen passes as preserving all analysis. 2008-09-22 20:58:04 +00:00
StrongPHIElimination.cpp Significant improvements to the logic for merging live intervals. This code can't 2008-09-22 21:58:58 +00:00
TargetInstrInfoImpl.cpp Remove isImm(), isReg(), and friends, in favor of 2008-09-13 17:58:21 +00:00
TwoAddressInstructionPass.cpp Tidy up several unbeseeming casts from pointer to intptr_t. 2008-09-04 17:05:41 +00:00
UnreachableBlockElim.cpp Remove isImm(), isReg(), and friends, in favor of 2008-09-13 17:58:21 +00:00
VirtRegMap.cpp Propagate subreg index when promoting a load to a copy. 2008-09-11 01:02:12 +00:00
VirtRegMap.h Move #include to right place. 2008-06-04 09:16:33 +00:00

README.txt

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

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.

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

I think we should have a "hasSideEffects" flag (which is automatically set for
stuff that "isLoad" "isCall" etc), and the remat pass should eventually be able
to remat any instruction that has no side effects, if it can handle it and if
profitable.

For now, I'd suggest having the remat stuff work like this:

1. I need to spill/reload this thing.
2. Check to see if it has side effects.
3. Check to see if it is simple enough: e.g. it only has one register
destination and no register input.
4. If so, clone the instruction, do the xform, etc.

Advantages of this are:

1. the .td file describes the behavior of the instructions, not the way the
   algorithm should work.
2. as remat gets smarter in the future, we shouldn't have to be changing the .td
   files.
3. it is easier to explain what the flag means in the .td file, because you
   don't have to pull in the explanation of how the current remat algo works.

Some potential added complexities:

1. Some instructions have to be glued to it's predecessor or successor. All of
   the PC relative instructions and condition code setting instruction. We could
   mark them as hasSideEffects, but that's not quite right. PC relative loads
   from constantpools can be remat'ed, for example. But it requires more than
   just cloning the instruction. Some instructions can be remat'ed but it
   expands to more than one instruction. But allocator will have to make a
   decision.

4. As stated in 3, not as simple as cloning in some cases. The target will have
   to decide how to remat it. For example, an ARM 2-piece constant generation
   instruction is remat'ed as a load from constantpool.

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

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