predessors of exit blocks from tail merging
consideration.

llvm-svn: 52985

llvm-svn: 52984

llvm-svn: 52983

llvm-svn: 52982

Add a version of AsmPrinter::EOL that takes a const char* so that we don't have to do as many implicit std::string constructions.
Unfortunately, this doesn't appear to translate to a real speedup in practice.

llvm-svn: 52981

llvm-svn: 52980

Thanks for the feedback!

llvm-svn: 52978

llvm-svn: 52977

llvm-svn: 52976

llvm-svn: 52975

llvm-svn: 52974

llvm-svn: 52972

llvm-svn: 52971

llvm-svn: 52970

llvm-svn: 52967

and mark it const along with the associated changes to TargetRegisterInfo.

llvm-svn: 52966

llvm-svn: 52963

llvm-svn: 52960

llvm-svn: 52957

llvm-svn: 52956

being suppressed here.

llvm-svn: 52952

  - Use a more accurate heuristic for the size of the hashtable.
  - Use bitwise and instead of modulo since the size is a power of two.
  - Use new[] instead of malloc().

llvm-svn: 52951

llvm-svn: 52950

llvm-svn: 52949

This fixes PR2503, though we should also fix other passes not to emit this kind of code.

llvm-svn: 52946

version that is computed by tblgen at the time LLVM is compiled.

llvm-svn: 52945

llvm-svn: 52944

the need for a flavor operand, and add a new SDNode subclass,
LabelSDNode, for use with them to eliminate the need for a label id
operand.

Change instruction selection to let these label nodes through
unmodified instead of creating copies of them. Teach the MachineInstr
emitter how to emit a MachineInstr directly from an ISD label node.

This avoids the need for allocating SDNodes for the label id and
flavor value, as well as SDNodes for each of the post-isel label,
label id, and label flavor.

llvm-svn: 52943

llvm-svn: 52940

Adding some missing files to VStudio project files.  Switching to use generated llvmAsmParser.cpp/.h

llvm-svn: 52935

llvm-svn: 52934

llvm-svn: 52933

llvm-svn: 52932

llvm-svn: 52931

llvm-svn: 52930

llvm-svn: 52929

instead of std::uninitialized_copy, which uses memmove.

llvm-svn: 52928

llvm-svn: 52927

SelectionDAG::allnodes_size is linear, but that doesn't appear to
outweigh the benefit of reducing heap traffic. If it does become a
problem, we should teach SelectionDAG to keep a count of how many
nodes are live, because there are several other places where that
information would be useful as well.

llvm-svn: 52926

llvm-svn: 52925