superceded by SplitKit.

llvm-svn: 118754

these points.

llvm-svn: 118752

it, so that it doesn't appear to be a known size.

llvm-svn: 118748

This is the first small step towards using closed intervals for liveness instead
of the half-open intervals we're using now.

We want to be able to distinguish between a SlotIndex that represents a variable
being live-out of a basic block, and an index representing a variable live-in to
its successor.

That requires two separate indexes between blocks. One for live-outs and one for
live-ins.

With this change, getMBBEndIdx(MBB).getPrevSlot() becomes stable so it stays
greater than any instructions inserted at the end of MBB.

llvm-svn: 118747

llvm-svn: 118746

llvm-svn: 118743

llvm-svn: 118742

constant loads.

llvm-svn: 118741

encoding bits.

llvm-svn: 118738

llvm-svn: 118736

llvm-svn: 118733

llvm-svn: 118732

llvm-svn: 118730

the reverse map too. This fixes seflhost build errors.

llvm-svn: 118729

llvm-svn: 118728

Take care of special characters while creating named MDNode name to hold function specific local variable's info.
This fixes radar 8653152. I am checking in testcase as a separate check-in.

llvm-svn: 118726

llvm-svn: 118725

for a given instruction into a helper function.

llvm-svn: 118723

llvm-svn: 118722

type is insufficient for, or incompatible with, the current query.

llvm-svn: 118721

llvm-svn: 118720

Probably it should just be 1, but compromise with 3.

llvm-svn: 118718

references. For example, this allows gvn to eliminate the load in
this example:

  void foo(int n, int* p, int *q) {
    p[0] = 0;
    p[1] = 1;
    if (n) {
      *q = p[0];
    }
  }

llvm-svn: 118714

llvm-svn: 118709

Whenever splitting wants to insert a copy, it checks if the value can be
rematerialized cheaply instead.

Missing features:
- Delete instructions when all uses have been rematerialized.
- Truncate live ranges to the remaining uses after rematerialization.

llvm-svn: 118702

benchmarks hitting an assertion.
Adds LiveIntervalUnion::collectInterferingVRegs.
Fixes "late spilling" by checking for any unspillable live vregs among
all physReg aliases.

llvm-svn: 118701

llvm-svn: 118699

instructions instead of hard-coding operand numbers.

llvm-svn: 118698

their comments.

llvm-svn: 118696

nodes can be used in loops, this could result in infinite looping
if there is no recursion limit, so add such a limit.  It is also
used for the SelectInst case because in theory there could be an
infinite loop there too if the basic block is unreachable.

llvm-svn: 118694

llvm-svn: 118693

it, and to be consistent.

llvm-svn: 118692

double quoting of ObjC symbol names in constant pool entries.

rdar://8652107

llvm-svn: 118688

arbitrary memory into a helper function, and adjust some comments.

llvm-svn: 118687

indivudal members holding the same data, to clarify the relationship
between NonLocalDepResult and NonLocalDepEntry.

llvm-svn: 118686

Store the flags needed to disable optimizations and to emit LLVM-IR depending on
the version of llvm-gcc used.

llvm-svn: 118684

llvm-svn: 118683

llvm-svn: 118682

llvm-svn: 118681

The simplifications performed here never create new instructions, they
only return existing instructions (or a constant), and so are always a
win.  In theory they should transform (for example)
  %z = and i32 %x, %y
  %s = select i1 %cond, i32 %y, i32 %z
  %r = and i32 %x, %s
into
  %r = and i32 %x, y
but in practice they get into a fight with instcombine, and lose.
Unfortunately instcombine does a poor job in this case.  Nonetheless
I'm committing this transform to make it easier to discuss what to
do to make peace with instcombine.

llvm-svn: 118679