instead of calling lower_bound or upper_bound directly.

This cleans up the search logic a bit because {lower,upper}_bound compare
LR->start by default, and it is usually simpler to search LR->end.

Funnelling all searches through one function also makes it possible to replace
the search algorithm with something faster than binary search.

llvm-svn: 114448

llvm-svn: 114447

SelectionDAG::getExtLoad overload, and eliminate it.

llvm-svn: 114446

getLoad overloads.

llvm-svn: 114443

with SVOffset computation.

llvm-svn: 114442

llvm-svn: 114437

no functionality change (step #1)

llvm-svn: 114436

llvm-svn: 114431

Added an additional PBQP problem builder which adds coalescing costs (both between pairs of virtuals, and between virtuals and physicals).

llvm-svn: 114429

into OptimizeCompareInstr.
This necessitates the passing of CmpValue around,
so widen the virtual functions to accomodate.

No functionality changes.

llvm-svn: 114428

pass a completely incorrect SrcValue, which would result in a miscompile with
combiner-aa.

llvm-svn: 114411

llvm-svn: 114409

"getFixedStack" on the MachinePointerInfo class.  While
this isn't the problem I'm setting out to solve, it is the
right way to eliminate PseudoSourceValue, so lets go with it.

llvm-svn: 114406

instead of srcvalue/offset pairs.  This corrects SV info for mem 
operations whose size is > 32-bits.

llvm-svn: 114401

llvm-svn: 114400

MachinePointerInfo.  Among other virtues, this doesn't silently  truncate the
svoffset to 32-bits.

llvm-svn: 114399

llvm-svn: 114397

eliminating some weird "infer a frame address" logic which was dead.

llvm-svn: 114396

llvm-svn: 114395

MachinePointerInfo, propagating the type out a level of API.  Remove
the old MachineFunction::getMachineMemOperand impl.

llvm-svn: 114393

to the MachineFunction construction methods.

llvm-svn: 114390

MachinePointerInfo struct, no functionality change.

This also adds an assert to MachineMemOperand::MachineMemOperand
that verifies that the Value* is either null or is an IR pointer type.

llvm-svn: 114389

define double @foo(double %x, double %y, i1 %c) nounwind {
  %a = fdiv double %x, 3.2
  %z = select i1 %c, double %a, double %y
  ret double %z
}

Was:
_foo:
        divsd   LCPI0_0(%rip), %xmm0
        testb   $1, %dil
        jne     LBB0_2
        movaps  %xmm1, %xmm0
LBB0_2:
        ret

Now:
_foo:
        testb   $1, %dil
        je      LBB0_2
        divsd   LCPI0_0(%rip), %xmm0
        ret
LBB0_2:
        movaps  %xmm1, %xmm0
        ret

This avoids the divsd when early exit is taken.
rdar://8454886

llvm-svn: 114372

CombinerAA cannot assume that different FrameIndex's never alias, but can instead use
MachineFrameInfo to get the actual offsets of these slots and check for actual aliasing.

This fixes CodeGen/X86/2010-02-19-TailCallRetAddrBug.ll and CodeGen/X86/tailcallstack64.ll
when CombinerAA is enabled, modulo a different register allocation sequence.

llvm-svn: 114348

llvm-svn: 114338

llvm-svn: 114313

r114268 fixed the last of the blockers to enabling it.  I will be monitoring
for failures.

llvm-svn: 114312

llvm-svn: 114284

llvm-svn: 114273

Added a separate class (PBQPBuilder) for PBQP Problem construction. This class can be extended to support custom constraints.

For now the allocator still uses the old (internal) construction mechanism by default. This will be phased out soon assuming 
no issues with the builder system come up.

To invoke the new construction mechanism just pass '-regalloc=pbqp -pbqp-builder' to llc. To provide custom constraints a
Target just needs to extend PBQPBuilder and pass an instance of their derived builder to the RegAllocPBQP constructor.

llvm-svn: 114272

llvm-svn: 114270

NO path to the destination containing side effects, not that SOME path contains no side effects.
In  practice, this only manifests with CombinerAA enabled, because otherwise the chain has little
to no branching, so "any" is effectively equivalent to "all".

llvm-svn: 114268

1) Do forward copy propagation. This makes it easier to estimate the cost of the
   instruction being sunk.
2) Break critical edges on demand, including cases where the value is used by
   PHI nodes.
Critical edge splitting is not yet enabled by default.

llvm-svn: 114227

llvm-svn: 114222

llvm-svn: 114220

If FE forgot to provide a file name (usually it uses "stdin" as name in such situation) then make one up to ensure that debug info is not malformed.

llvm-svn: 114119

great deal because we don't have to worry about maintaining SSA form.

Unconditionally copy back to dupli when the register is live out of the split
range, even if the live-out value was defined outside the range. Skipping the
back-copy only makes sense when the live range is going to spill outside the
split range, and we don't know that it will. Besides, this was a hack to avoid
SSA update issues.

Clear up some confusion about the end point of a half-open LiveRange. Methinks
LiveRanges need to be closed so both start and end are included in the range.
The low bits of a SlotIndex are symbolic, so a half-open range doesn't really
make sense. This would be a pervasive change, though.

llvm-svn: 114043

This fixes funcargs.exp regression reported by gdb testsuite.

llvm-svn: 113992

If dbg.declare from non-entry block is using alloca from entry block then use offset available in StaticAllocaMap to emit DBG_VALUE. Right now, this has no material impact because varible info also collected using offset table maintained in machine module info.

llvm-svn: 113967

llvm-svn: 113848