deduction using a base class of the argument type. No actual
functionality change; this is just a hook.

llvm-svn: 74356

When possible, use "mvn ra, rb" instead of "eor ra, rb, -1" because mvn has a narrow version and eor(i) does not.

llvm-svn: 74355

deduction from pointer and pointer-to-member types to work even in the
presence of a qualification conversion (C++ [temp.deduct.type]p3
bullet 2). 

llvm-svn: 74354

llvm-svn: 74353

This helps it avoid reusing an instruction that doesn't dominate all
of the users, in cases where the original instruction was inserted
before all of the users were known.  This may result in redundant
expansions of sub-expressions that depend on loop-unpredictable values
in some cases, however this isn't very common, and it primarily impacts
IndVarSimplify, so GVN can be expected to clean these up.

This eliminates the need for IndVarSimplify's FixUsesBeforeDefs,
which fixes several bugs.

llvm-svn: 74352

llvm-svn: 74351

Add feature flags for AVX and FMA and fix some SSE4A feature flag
initialization problems.

llvm-svn: 74350

llvm-svn: 74345

nesting order of nested AddRec expressions to skip the transformation
if it would introduce an AddRec with operands not loop-invariant
with respect to its loop.

llvm-svn: 74343

instead of skipping).

llvm-svn: 74341

llvm-svn: 74340

llvm-svn: 74339

are loop invariant, not just the start operand.

llvm-svn: 74338

llvm-svn: 74337

Add x86 support for 'n' inline asm modifier. This will be handled target independently as part of MC work.

llvm-svn: 74336

llvm-svn: 74335

llvm-svn: 74334

llvm-svn: 74333

llvm-svn: 74332

of template instantiation, we were dropping cv-qualifiers on the
instantiated type in a few places. This change reshuffles the
type-instantiation code a little bit so that there's a single place
where we add qualifiers to the instantiated type, so that we won't end
up with this same bug in the future.

llvm-svn: 74331

llvm-svn: 74330

Step 1: ARMInstructionInfo => {ARM,Thumb}InstructionInfo

llvm-svn: 74329

llvm-svn: 74328

llvm-svn: 74327

llvm-svn: 74326

the SelectionDAG::getGlobalAddress function properly looks through
aliases to determine thread-localness, but then passes the GV* down
to GlobalAddressSDNode::GlobalAddressSDNode which does not.  Instead
of passing down isTarget, just pass down the predetermined node
opcode.  This fixes some assertions with out of tree changes I'm 
working on.

llvm-svn: 74325

llvm-svn: 74324

non-dependent parameter types. Instead, class template partial
specializations perform a final check of all of the instantiated
arguments. This model is cleaner, and works better for function
templates where the "final check" occurs during overload resolution.

Also, cope with cv-qualifiers when the parameter type was originally a
reference type, so that the deduced argument can be more qualified
than the transformed argument.

llvm-svn: 74323

llvm-svn: 74322

llvm-svn: 74321

llvm-svn: 74320

we had multiple type planes and these lookups were expensive.

llvm-svn: 74319

llvm-svn: 74317

llvm-svn: 74316

llvm-svn: 74315

llvm-svn: 74314

llvm-svn: 74313

llvm-svn: 74312

SDNode::print_details to eliminate a ton of near-duplicate code.

llvm-svn: 74311

llvm-svn: 74310