otherwise unhoistable instructions in order to allow the loads
to be hoisted.

llvm-svn: 85364

llvm-svn: 85363

llvm-svn: 85362

llvm-svn: 85361

MachineLICM and other things which run before LiveVariables is run.

llvm-svn: 85360

if they have compatible encodings.

llvm-svn: 85359

llvm-svn: 85358

eliminating a use of MVT::Flag, this is needed for an upcoming CodeGen
change.

This unfortunately requires SystemZ to switch to the list-burr
scheduler, in order to handle the physreg defs properly, however
that's what LLVM has available at this time.

llvm-svn: 85357

llvm-svn: 85355

llvm-svn: 85354

llvm-svn: 85353

llvm-svn: 85352

llvm-svn: 85351

llvm-svn: 85350

untested and there is no way to use it, next up: doing battle
with asmparser.

llvm-svn: 85349

explicit-instantiation-declaration-after-explicit-instantiation-definition
errors. This wraps up explicit template instantiation for now.

llvm-svn: 85347

use it to control tail merging when there is a tradeoff between performance
and code size.  When there is only 1 instruction in the common tail, we have
been merging.  That can be good for code size but is a definite loss for
performance.  Now we will avoid tail merging in that case when the
optimization level is "Aggressive", i.e., "-O3".  Radar 7338114.

Since the IfConversion pass invokes BranchFolding, it too needs to know
the optimization level.  Note that I removed the RegisterPass instantiation
for IfConversion because it required a default constructor.  If someone
wants to keep that for some reason, we can add a default constructor with
a hard-wired optimization level.

llvm-svn: 85346

llvm-svn: 85345

colliding with llvm/ADT/ValueMap.h:ValueMap<>.

llvm-svn: 85344

llvm-svn: 85343

llvm-svn: 85341

inlined functions. For example, given

  template<typename T>
  class string {
    unsigned Len;

  public:
    unsigned size() const { return Len; }
  };

  extern template class string<char>;

we now give the instantiation of string<char>::size
available_externally linkage (if it is ever instantiated!), as
permitted by the C++0x standard.
      

llvm-svn: 85340

llvm-svn: 85339

Fixed a bug in the coalescer where intervals were occasionally merged despite a real interference. This fixes rdar://problem/7157961.

llvm-svn: 85338

class pointer is the most derived common class of the two.
This is <rdar://problem/7334235>.

llvm-svn: 85337

llvm-svn: 85335

llvm-svn: 85334

default behind a command line option. This will enable better performance for
vectors on NEON enabled processors.

llvm-svn: 85333

llvm-svn: 85332

llvm-svn: 85331

against both 2.6 and HEAD.  The default is still changed to eager jitting.

llvm-svn: 85330

llvm-svn: 85329

llvm-svn: 85327

llvm-svn: 85325

side-effects up front, as when we switch to the llvm intrinsic call
for __builtin_object_size later, it will have two evaluations.

We also finish off the intrinsic version of the code so we can just
turn it on once llvm has the intrinsic.

llvm-svn: 85324

llvm-svn: 85323

llvm-svn: 85322

llvm-svn: 85321

llvm-svn: 85320

llvm-svn: 85319