ScheduleDAGEmit, TwoAddressLowering, and PHIElimination.

This switches the bulk of register copies to using COPY, but many less used
copyRegToReg calls remain.

llvm-svn: 108050

llvm-svn: 108048

Based on a patch by Rafael Espíndola.

Attempt to make the FpSET_ST1 hack more robust, but we are still relying on
FpSET_ST0 preceeding it. This is only for supporting really weird x87 inline
asm.

We support:

  FpSET_ST0
  INLINEASM

  FpSET_ST0
  FpSET_ST1
  INLINEASM

with and without kills on the arguments. We don't support:

  FpSET_ST1
  FpSET_ST0
  INLINEASM

nor

  FpSET_ST1
  INLINEASM

Just Don't Do It!

llvm-svn: 108047

llvm-svn: 108046

at the end of the block.

llvm-svn: 108045

default arguments to template parameters don't have a DeclContext when
instantiated, and so we can't detect that we're in an instantiation context as
opposed to the definition context. However, it fixes the more commonly-occuring
cases in TMP code that use devolve to this type of tautology after
substitution.

llvm-svn: 108044

llvm-svn: 108043

llvm-svn: 108042

llvm-svn: 108041

llvm-svn: 108040

 - Check getBytesToPopOnReturn().
 - Eschew ST0 and ST1 for return values.
 - Fix the PIC base register initialization so that it doesn't ever
   fail to end up the top of the entry block.

llvm-svn: 108039

it is popped, even if it is ununsed. A CopyFromReg node is too weak to represent
the required sideeffect, so insert an FpGET_ST0 instruction directly instead.

This will matter when CopyFromReg gets lowered to a generic COPY instruction.

llvm-svn: 108037

This relies on ThreadPlanStepOut working correctly, which it doesn't currently for Inlined functions, so this feature is only partially useful until we take care of Stepping Out of inlined functions.
Added an option to "thread step-in" to set the avoid regular expression.  This is mostly for testing, once the Setting code is redone, we'll move this to a general setting.

llvm-svn: 108036

llvm-svn: 108035

llvm-svn: 108031

llvm-svn: 108030

llvm-svn: 108029

llvm-svn: 108028

a bunch of BSD-isms.

llvm-svn: 108027

be a dependent expression when its is built.

llvm-svn: 108026

tell me if this breaks anything.

llvm-svn: 108025

llvm-svn: 108024

llvm-svn: 108023

llvm-svn: 108022

llvm-svn: 108021

llvm-svn: 108020

block literal expression.

llvm-svn: 108019

Slightly improve the diagnostic when using a qualified function typedef to declare nonmember or static member functions.

llvm-svn: 108018

llvm-svn: 108017

possibility of adding an unitialized one into the folding set.

llvm-svn: 108016

llvm-svn: 108015

When given the -chained-pch option and a previous PCH file, have the PCHWriter emit a CHAINED_METADATA record instead of METADATA, and write a link to the previous file there.

llvm-svn: 108014

llvm-svn: 108013

llvm-svn: 108012

llvm-svn: 108011

llvm-svn: 108010

enabled LLVM make style building and made this compile LLDB on Mac OS X. We
can now iterate on this to make the build work on both linux and macosx.

llvm-svn: 108009

rdar://8131327

llvm-svn: 108008

for code like this:
   template<template<typename T> class U> class V {};

The problem is that the DeclPrinter assumed all TemplateDecls
have a getTemplatedClass(), but template template params don't
(so we got a NULL dereference).  The solution is to detect if
we're a template template param, and construct the template
class name ('class U') specially in this case.

OKed by dgregor and chandlerc

llvm-svn: 108007

llvm-svn: 108006