llvm-svn: 72729

entries as there are basic blocks in the function.  LiveVariables::getVarInfo
creates a VarInfo struct for every register in the function, leading to
quadratic space use.  This patch changes the BitVector to a SparseBitVector,
which doesn't help the worst-case memory use but does reduce the actual use in
very long functions with short-lived variables.

llvm-svn: 72426

In some rare cases, the register allocator can spill registers but end up not utilizing registers at all. The fundamental problem is linearscan's backtracking can end up freeing more than one allocated registers. However,  reloads and restores might be folded into uses / defs and freed registers might not be used at all.

VirtRegMap keeps track of allocations so it knows what's not used. As a horrible hack, the stack coloring can color spill slots with *free* registers. That is, it replace reload and spills with copies from and to the free register. It unfold instructions that load and store the spill slot and replace them with register using variants.

Not yet enabled. This is part 1. More coming.

llvm-svn: 70787

llvm-svn: 70245

Fix PR4056. It's possible a physical register def is dead if its implicit use is deleted by two-address pass.

llvm-svn: 70213

This fixes a very subtle bug. vr defined by an implicit_def is allowed overlap with any register since it doesn't actually modify anything. However, if it's used as a two-address use, its live range can be extended and it can be spilled. The spiller must take care not to emit a reload for the vn number that's defined by the implicit_def. This is both a correctness and performance issue.

llvm-svn: 69743

Added a linearscan register allocation optimization. When the register allocator spill an interval with multiple uses in the same basic block, it creates a different virtual register for each of the reloads. e.g.

	%reg1498<def> = MOV32rm %reg1024, 1, %reg0, 12, %reg0, Mem:LD(4,4) [sunkaddr39 + 0]
        %reg1506<def> = MOV32rm %reg1024, 1, %reg0, 8, %reg0, Mem:LD(4,4) [sunkaddr42 + 0]
        %reg1486<def> = MOV32rr %reg1506
        %reg1486<def> = XOR32rr %reg1486, %reg1498, %EFLAGS<imp-def,dead>
        %reg1510<def> = MOV32rm %reg1024, 1, %reg0, 4, %reg0, Mem:LD(4,4) [sunkaddr45 + 0]

=>

        %reg1498<def> = MOV32rm %reg2036, 1, %reg0, 12, %reg0, Mem:LD(4,4) [sunkaddr39 + 0]
        %reg1506<def> = MOV32rm %reg2037, 1, %reg0, 8, %reg0, Mem:LD(4,4) [sunkaddr42 + 0]
        %reg1486<def> = MOV32rr %reg1506
        %reg1486<def> = XOR32rr %reg1486, %reg1498, %EFLAGS<imp-def,dead>
        %reg1510<def> = MOV32rm %reg2038, 1, %reg0, 4, %reg0, Mem:LD(4,4) [sunkaddr45 + 0]

From linearscan's point of view, each of reg2036, 2037, and 2038 are separate registers, each is "killed" after a single use. The reloaded register is available and it's often clobbered right away. e.g. In thise case reg1498 is allocated EAX while reg2036 is allocated RAX. This means we end up with multiple reloads from the same stack slot in the same basic block.

Now linearscan recognize there are other reloads from same SS in the same BB. So it'll "downgrade" RAX (and its aliases) after reg2036 is allocated until the next reload (reg2037) is done. This greatly increase the likihood reloads from SS are reused.

This speeds up sha1 from OpenSSL by 5.8%. It is also an across the board win for SPEC2000 and 2006.

llvm-svn: 69585

a live interval. This is needed for some upcoming subreg changes.

llvm-svn: 68956

register destinations that are tied to source operands.  The
TargetInstrDescr::findTiedToSrcOperand method silently fails for inline
assembly.  The existing MachineInstr::isRegReDefinedByTwoAddr was very
close to doing what is needed, so this revision makes a few changes to
that method and also renames it to isRegTiedToUseOperand (for consistency
with the very similar isRegTiedToDefOperand and because it handles both
two-address instructions and inline assembly with tied registers).

llvm-svn: 68714

with SUBREG_TO_REG, teach SimpleRegisterCoalescing to coalesce
SUBREG_TO_REG instructions (which are similar to INSERT_SUBREG
instructions), and teach the DAGCombiner to take advantage of this on
targets which support it. This eliminates many redundant
zero-extension operations on x86-64.

This adds a new TargetLowering hook, isZExtFree. It's similar to
isTruncateFree, except it only applies to actual definitions, and not
no-op truncates which may not zero the high bits.

Also, this adds a new optimization to SimplifyDemandedBits: transform
operations like x+y into (zext (add (trunc x), (trunc y))) on targets
where all the casts are no-ops. In contexts where the high part of the
add is explicitly masked off, this allows the mask operation to be
eliminated. Fix the DAGCombiner to avoid undoing these transformations
to eliminate casts on targets where the casts are no-ops.

Also, this adds a new two-address lowering heuristic. Since
two-address lowering runs before coalescing, it helps to be able to
look through copies when deciding whether commuting and/or
three-address conversion are profitable.

Also, fix a bug in LiveInterval::MergeInClobberRanges. It didn't handle
the case that a clobber range extended both before and beyond an
existing live range. In that case, multiple live ranges need to be
added. This was exposed by the new subreg coalescing code.

Remove 2008-05-06-SpillerBug.ll. It was bugpoint-reduced, and the
spiller behavior it was looking for no longer occurrs with the new
instruction selection.

llvm-svn: 68576

llvm-svn: 67764

llvm-svn: 67544

Model inline asm constraint which ties an input to an output register as machine operand TIED_TO constraint. This eliminated the need to pre-allocate registers for these. This also allows register allocator can eliminate the unneeded copies.

llvm-svn: 67512

 - Make type declarations match the struct/class keyword of the definition.
 - Move AddSignalHandler into the namespace where it belongs.
 - Correctly call functions from template base.
 - Some other small changes.
With this patch, LLVM and Clang should build properly and with far less noise under VS2008.

llvm-svn: 67347

llvm-svn: 67335

Fix how livein live intervals are handled. Previously it could end at MBB start. Sorry, no small test case possible.

llvm-svn: 66129

Fix PR3486. Fix a bug in code that manually patch physical register live interval after its sub-register is coalesced with a virtual register.

llvm-svn: 64082

llvm-svn: 63267

llvm-svn: 62600

The coalescer does not coalesce a virtual register to a physical register if any of the physical register's sub-register live intervals overlaps with the virtual register. This is overly conservative. It prevents a extract_subreg from being coalesced away:

v1024 = EDI  // not killed
      =
      = EDI

One possible solution is for the coalescer to examine the sub-register live intervals in the same manner as the physical register. Another possibility is to examine defs and uses (when needed) of sub-registers. Both solutions are too expensive. For now, look for "short virtual intervals" and scan instructions to look for conflict instead.

This is a small win on x86-64. e.g. It shaves 403.gcc by ~80 instructions.

llvm-svn: 61847

Fix PR3149. If an early clobber def is a physical register and it is tied to an input operand, it effectively extends the live range of the physical register. Currently we do not have a good way to represent this.

172     %ECX<def> = MOV32rr %reg1039<kill>
180     INLINEASM <es:subl $5,$1
        sbbl $3,$0>, 10, %EAX<def>, 14, %ECX<earlyclobber,def>, 9, %EAX<kill>,
36, <fi#0>, 1, %reg0, 0, 9, %ECX<kill>, 36, <fi#1>, 1, %reg0, 0
188     %EAX<def> = MOV32rr %EAX<kill>
196     %ECX<def> = MOV32rr %ECX<kill>
204     %ECX<def> = MOV32rr %ECX<kill>
212     %EAX<def> = MOV32rr %EAX<kill>
220     %EAX<def> = MOV32rr %EAX
228     %reg1039<def> = MOV32rr %ECX<kill>

The early clobber operand ties ECX input to the ECX def.

The live interval of ECX is represented as this:
%reg20,inf = [46,47:1)[174,230:0)  0@174-(230) 1@46-(47)

The right way to represent this is something like
%reg20,inf = [46,47:2)[174,182:1)[181:230:0)  0@174-(182) 1@181-230 @2@46-(47)

Of course that won't work since that means overlapping live ranges defined by two val#.

The workaround for now is to add a bit to val# which says the val# is redefined by a early clobber def somewhere. This prevents the move at 228 from being optimized away by SimpleRegisterCoalescing::AdjustCopiesBackFrom.

llvm-svn: 61259

llvm-svn: 60683

Reason #3 from 60595 doesn't hold true. If we can fold a PIC load from constpool into a use, the rewrite happens at time of spill (not in VirtRegMap). Later on, if the GlobalBaseReg is spilled, the spiller can see the use uses GlobalBaseReg and do the right thing.

llvm-svn: 60596

llvm-svn: 60592

llvm-svn: 60586

llvm-svn: 60487

and the LiveInterval.h top-level comment and accordingly. This fixes blocks
having spurious live-in registers in boundary cases.

llvm-svn: 60092

llvm-svn: 59841

BitVector, instead of manually testing each bit.

llvm-svn: 59246

coalescing as a separate pass rather than inside of
LiveIntervalAnalysis.

llvm-svn: 59146

- Create and update spill slot live intervals.
- Lots of bug fixes.

llvm-svn: 58367

Fix PR2634.  Create new virtual registers from spills early so that we
can give it the same stack slot as the spilled interval if it is folded.
This prevents the fold/unfold code from pointing to the wrong register.

llvm-svn: 58255

llvm-svn: 58068

llvm-svn: 57766

llvm-svn: 57765

llvm-svn: 57259

isReg, etc., from isRegister, etc.

llvm-svn: 57006

Fix a simple error in renumbering kill markaers, that took an inordinant amount of time to track down.

llvm-svn: 56889

llvm-svn: 56848

"If a re-materializable instruction has a register
operand, the spiller will change the register operand's
spill weight to HUGE_VAL to avoid it being spilled.
However, if the operand is already in the queue ready
to be spilled, avoid re-materializing it".

llvm-svn: 56837